General5
NoteUpdate 1 (October 1): Dispense Complete and Emergency Stop pressed on a web page reach the dispensing hardware again, and so do the stop of a reel calibration and setting the controller's time.
NoteUpdate 2 (October 1): a Dispense Complete that measured no fluid asks whether to keep the line open or close it; the reel-calibration window keeps its pulse count; a notice appears when every radio device has stopped answering for two minutes; the Communication Module is always treated as wired.
NoteUpdate 3 (October 2): station, product and user names with accents no longer stop dispenses from completing; report e-mails, CSV downloads and FTP exports keep accented names; the Emergency Stop, over-dispense and overtime notices appear on every page; after an Emergency Stop the amount offered to finish a line is exact, so the line cannot be overfilled; a tank without a level sensor is lowered by every dispense that moved oil, including one ended by an Emergency Stop.
NoteUpdate 4 (October 3): every Action Journal entry records its real date and time, and older entries are corrected by the update; the transfer to OilCop support no longer stops when a single record is refused; a dispense is booked to the meter's final count (some were booked one display step short); messages from the dispensing hardware are no longer occasionally discarded; the Close Stale Work Orders button works again; the Transaction Journal CSV dates evening dispenses correctly and report date stamps follow the date format set on the controller; a disabled manager's PIN no longer clears an Emergency Stop or approves a work order deletion; adding an installer requires LiquiDynamics' key PIN; keypad PINs are no longer written to the logs; an uploaded import file is deleted once it has been read.
NoteUpdate 5 (October 4): the transfer to OilCop support connects from networks that answer slowly, always uses the controller's own name at OilCop support, and no longer has user records or Action Journal entries refused; changes to devices, deleted dispenses and deleted work orders reach OilCop support every time, a record changed during a transfer is sent again, and values OilCop support used to shorten (Action Journal texts, reel calibration factors, tank levels, the logo file name) are sent once more in full; Action Journal entries of up to 1,000 characters are kept whole (each install used to shorten them); the update e-mail shows whether the controller's transfer settings are as expected; the hourly health report identifies the controller's own disk, so two controllers that share one network address (MAC) can be told apart. Also in Update 5: the update pages can no longer reinstall from a refreshed, restored or old browser tab (each install link works once, and the page changes its own address when the install starts); an update file that is an older build of the installed version is installed only after a red warning and a second button press; when the stored update file is gone, the page says so instead of downloading another one; Install from a file shows the installed build next to the file's build; the update e-mail no longer says "installed from a file" after a download; an "All Products" line on a work order appears once every listed product is finished (the manager's preset or the technician's preset reached, an open dispense, or Dispense Complete pressed); the Apply button of Administrative Options stays in view while the page scrolls. For support: an install address typed by hand (...&doUpdate=true) no longer installs -- use Update System > Check for Specific Version > Use Existing ZIP > Install Now.
Dispensing29
NewApproval of dispenses over a limit. When a technician asks for more than a cap allows, the request goes to a manager, showing how far over the cap it is and how much the work order has already dispensed. Every cap works this way: the preset limit, the product's per-work-order limit and the work order's total limit. The pending request appears at the top of every page an administrator or manager is on. One approval covers the caps in play; it arms the Start button and the technician still presses Start. The most recent manager decision is the one that counts, and a denial is withdrawn when the technician changes the amount.
NewThe dispense monitor. An administrator or manager clicks an active reel on the dashboard to watch the technician's dispense live (amount, work order, product, station, preset). It only watches and cannot affect it.
NewShift hours can also be enforced at the pump. Two Administrative Options, "Enforce Shift Hours at the Pump" and "Include Managers in Shift Hours", are both OFF by default; off, a shift limits the login only, as before. On, a technician cannot press Start outside their shift even if already logged in. A technician refused at login is told the shift's name and hours, and an off-shift technician can finish a dispense under way.
NewApproved overtime. A technician refused outside their shift, at the login screen, the Start button or the keypad (TAC), can request overtime; administrators and managers approve it on their dashboard, for that technician or all technicians and for a chosen time, and can end it early. Overtime always ends. At the keypad the refusal itself raises the request, and the next PIN is accepted once it is approved. Every request, approval, cancellation and overtime login is in the System Log.
ChangedWhen Top Off is enabled it is offered after every completed preset dispense, whatever the product's per-work-order limit stands at. The limit is confirmed when the technician presses Start, and a top-off that would go over it gets the manager approval offer, like any other dispense.
ChangedThe technician sees the whole work order they are working on, and what each product has already taken, on the dispense page.
ChangedAt the keypad (TAC), the work order's Lock Dispense Amount box decides whether a manager's preset binds, as on the web and tablet pages. Locked, a line with a manager's preset is not asked for an amount and pumps the manager's preset; unlocked, the keypad asks, and the amount typed is the amount pumped. A line without a manager's preset is asked for, as before.
ChangedSaving Administrative Options writes one System Log line per option that actually changed ("Odometer Number OFF to ON"), under the name of who saved; a first-time choice reads "(not set)".
ImprovedWhen Start is not allowed, the page says why (no amount entered, no reel assigned, a refused start), and dispensing errors appear as a normal OilCop message with a way back instead of a line of text on a blank page.
FixedOn Gartec and Admiral/Filcar controllers every dispense record carried the same user. Each dispense now records the technician who made it.
FixedMultiple Products and Multiple Dispense work orders follow the Administrative Options. A manager's work order with both switched off failed silently at the pump, a Multiple Products work order that dispensed nothing was closed automatically, and returning to a work order after a top-off or on an All Products line did not always pick up where it left off.
FixedThe web dispense page could leave duplicate "phantom" dispense rows.
FixedA tank set up without a unit could block all dispensing with a "not enough fluid" message.
FixedA technician's preset was overwritten with the amount actually delivered once the dispense finished, so preset and delivered amounts looked identical on reports, showed odd values such as 4.09615, and a dispense that came up short could not be resumed properly.
FixedStarting a job dispense at a station with more than one reel for the same product ended with "Reel is not found." and no way forward.
FixedA top-off could hand out more oil than was left of its allowance. The allowance is now calculated on the controller from what the work order has already taken, and only what is left of it can be dispensed.
FixedA work order kept open after a top-off could be trapped by a stale top-off flag, and a top-off that timed out could not be closed automatically after its work order was closed.
FixedA preset stopped with the emergency stop could resume as "Preset 0". Every dispense now records the active preset and where it came from, so a resumed dispense shows what it is still owed, in the right units even when the reel and the cap use different ones.
FixedThe emergency stop is aimed at the pump that is actually running, and no longer goes deaf for 22 seconds after the serial link is reopened.
FixedA set preset is bound to the dispense; the keyboard Enter key sets one, and a dispense resumes correctly after a PIN.
FixedOn the tablet the preset keypad rounded an entered amount to one decimal and dispensed the rounded figure (1.89 became 1.9). It now checks the amount as the desktop page does; both accept one decimal place.
FixedA dispense too small for the meter to measure was accepted and pumped nothing. It is now refused with a message.
FixedA dispense that ran longer than the Dispense Complete timeout could be closed while the pump was still running, so the rest of the preset was delivered but never recorded and tank inventory drifted. A dispense is now closed only when its amount has stopped changing for the whole timeout, and every such close is in the dispense audit log.
FixedA very small remainder at the end of a preset could start the pump for a few pulses three times over.
FixedJobs started from the keypad (TAC) pump the recipe at full precision, not rounded to the displayed tenths.
FixedA work-order receipt could print twice; it now prints once, when the work order is closed.
FixedOn the tablet the dispense page matches the desktop page, and the number keypad appears as a centred card without the tablet's own keyboard.
FixedA technician who signs out, or is signed out, is told why.
FixedThe Cancel button on the reel calibration screen did nothing. It now stops the pump if it is running and closes the calibration screen.
Work orders12
ChangedOn the web and tablet dispense pages, a work order that has reached a limit closes itself after the dispense and its top-off finish, instead of asking whether to keep it open. A limit reached on one product under Multiple Products leaves the others dispensable; the work order closes when its total limit is reached. Work orders under a limit ask keep open or close as before. At a keypad controller a work order that reaches its limit closes when Auto-Close Work Order is enabled, as before.
ChangedWith Multiple Dispense on, the Keep Open or Close Work Order prompt appears whenever a technician leaves the dispense screen, with Back or Dispense Complete, so any multi-dispense work order can be closed there.
ChangedClosing a work order needs a login, and a work order cannot be closed while someone is still dispensing on it.
ChangedWork orders are named by the customer's work order number, in messages and in the journal, rather than by an internal number.
ChangedThe technician's product list shows only a preset the manager set, with its unit and marked "(locked)" when locked, and Amount Dispensed carries its unit in the cell.
ChangedRecent Work Orders and Work Order Maintenance follow one rule: open work orders always show; closed ones follow the Hide Closed Work Orders window, by the time they were closed. The Invoice column is a Yes/No toggle and the status column has a working Open/Close toggle.
ImprovedThe Enter Work Order prompt puts the cursor in the field, submits on Enter, and clearly flags a repeated invalid entry.
ImprovedThe Maintenance page's Close Stale Work Orders and Cancelled Work Orders lists keep their action button in view, have readable headings, and can be printed or exported to CSV.
FixedTwo technicians can share a work order without sharing a line; a second technician can no longer walk onto a line someone is dispensing on, and any technician can finish a product a colleague started.
FixedA visit that dispensed nothing now ends with Dispense Complete instead of leaving the work order half open.
FixedA product dispensed from two reels measured in different units was totalled as if the units were the same (a quart plus a litre read as two litres). Totals and the top-off checks now work in one unit.
FixedWork Order Maintenance shows the correct top-off amount for each work order, and gallons are shown as "G".
Tanks, inventory and deliveries12
NewWith the deliveries option on, a delivery detected by a sensor tank can be e-mailed to the people subscribed to "Oil Delivery" in the Email Scheduler, with a link to its paperwork. The Oil Deliveries register hides sensor increases below the delivery minimum behind a "Show all" tick box.
NewIf tank adjustments ever stop being recorded, the Action Journal says so instead of going quiet about it.
ImprovedTank monitoring no longer records minor sensor fluctuations as inventory changes.
FixedChanging a tank's units while the system was running could record a false inventory adjustment.
FixedA tank with a custom strapping chart reported EMPTY when filled above the highest level in its chart, or when its sensor returned an invalid reading, and could raise a false low-level alert. It now reports the highest charted level and records a warning so the chart can be extended.
FixedAutomatic inventory entries in the Action Journal now say which unit the amount is in, like entries made by hand.
FixedOne tank returning bad readings could freeze every tank on the same module; the other tanks now keep reporting.
FixedWhen the controller loses contact with the module, the devices on it are shown offline instead of keeping their last state.
FixedOil dispensed from a tank without a level sensor is now subtracted from that tank's inventory.
FixedTank-monitor dispensing refuses an ambiguous relay setup and always enforces the relay timeout.
FixedThe Add Inventory and Subtract Inventory windows opened from Tank Settings showed the tank's figures under the wrong labels.
FixedA tank whose name used non-English letters could lose a delivery.
Reports and email18
NewThe Work Order Status report has a Cancelled Work Orders section, and completed work orders are grouped by the date they were closed.
NewThe E-mail Server page (was SMTP Settings) works with a customer's own mail server in every common shape: the branded server (default), a relay on your own network with no login, or a mail account with a login (Office 365, Google, hosted). One Connection choice (None port 25, STARTTLS port 587, SSL/TLS port 465) replaces the SSL and TLS switches; a separate Login name covers shared mailboxes and service accounts; "Send test e-mail" tries the values before you save and shows the mail server's own reply. Going back to the branded server keeps your server for later, and every change is in the System Log. Every port other than 587 used to send nothing at all.
NewAn e-mail that cannot be sent is recorded in the Action Journal with a plain-words reason (for example that the mail server rejected the login), repeated failures only once, so a mail problem is visible on the controller.
ChangedThe Email Scheduler has been reworked: a grid of reports and days, a time chooser in half-hour steps, Weekly and Monthly on every report, clearer names ("High Level Shutoff", "Low Level Shutoff", "Summary Report"), an information icon on each report explaining when it sends, and external recipients who are not system users.
ChangedEvery e-mail the controller sends has one layout: the message first and the logo at the end, at the same size in every mail program. A report e-mail says which report is attached and for which site.
ChangedThe Tank Adjustments, Action Journal, Transaction Journal and Work Order Status PDFs are landscape, so the wider columns fit.
ChangedEntries the system records by itself, such as automatic inventory changes and the daily close of stale work orders, are shown as "System". They were recorded against the administrator account, which named a real person, and on Gartec or Admiral systems the wrong company's account.
ImprovedThe System Log shows people's names instead of internal numbers, and entries whose author cannot be matched are shown instead of left out.
ChangedSuccess messages are green and close by themselves; failures are red and stay on screen until closed.
FixedA Daily Report for a past date described today in several places: the Current Inventory figure and tank bar, the Consumption Trends windows and the Work Order section with its Opened / Total Open / Closed counts. All of them now describe the reported date, and the headings say so. The 14 day Average shows "At/below reorder" instead of a dash for a tank that needs reordering.
FixedThe Daily report's work-order listing and heading were about five hours early. The report is built once, and the scheduled e-mail is sent from that same report.
FixedOn the PDF Daily Report the last group of tanks matches the groups above it, and its headings are translated.
FixedHigh-level and low-level shutoff alert e-mails could fail to send; their subjects now say which alarm it is.
FixedThe "Send Daily Report" button on the Email Schedule page did nothing, while showing a message saying the mail had been sent.
ChangedThe page footer of every PDF report ("Page 1 / 3") is in the controller's language.
FixedAn empty or unreadable logo file stopped the controller's report and scheduled e-mails. The e-mail now goes with the brand's logo, or without one.
FixedOn Admiral/Filcar controllers the update and hardware alert e-mails arrived with an empty body; they now carry their text.
FixedThe schedule popup on the Email Schedule page could show one report's saved schedule against another report; each report now shows its own.
Users, access and screens18
NewReport Issue replaces the Help button in the header of every page. It opens a dialog on the page you are on, takes a picture of that page, asks what happened, can send you a copy, and e-mails the report to your brand's support team (LiquiDynamics, Gartec or Filcar) with the picture and a zip of what support needs to start: two weeks of Action Journal, dispenses and work orders, a current copy of the configuration with every PIN and password withheld, and the recent start-up, update and error logs. Each report gets a reference number, is kept on the controller, and the Reported Issues tab lists what has been sent. Where the controller's mail is blocked, the dialog gives up within seconds and offers the files to download with the address, subject and message to copy. The button can always be pressed, even over a notice or a startup screen. OilCop-branded controllers also show the link to OilCop Support at oilcopsupport.com there.
NewA brand selector, so e-mails and reports carry the right company's branding, and a Company Site Identifier that labels e-mail subjects and report headers, with a dashboard reminder while it is not set.
ChangedThe footer of every page shows the build of the installed software next to the version. The copyright line names LiquiDynamics on every brand, and the "powered by" name follows the brand.
ChangedThe administrator menu lists Work Order Maintenance, Adjust Job Recipe, Adjust Tank Inventory and Technician Jobs, so an administrator no longer switches to the Manager view to reach them.
ChangedManager menu items can be opened in a new tab like the administrator menu, and the reports menu is labelled System Reports.
ChangedAn uploaded logo keeps a readable file name, and choosing a header logo no longer erases the Company Name text.
ChangedAdministrative Options is shown in the controller's language; several of its labels were in English in every language.
NewThe controller's file manager has bookmarks for the folders OilCop support asks for (System logs, Report an Issue zips). Bookmarks added on the controller are kept.
ImprovedThe admin dashboard remembers which sections you collapsed.
ImprovedReboot and restart actions record which user performed them.
FixedA login could land the user on the Dispense Page instead of the dashboard, with an empty Switch-To user list.
FixedTechnician names with accented characters display correctly, and screen labels that were wrong, untranslated or shown by their internal name in some languages have been corrected.
FixedKeypad (TAC) logins for sites with many users. The keypad can only address 255 user numbers; users created beyond that could not log in, or logged in as someone else. User numbers now stay within the range, and any already beyond it are repaired automatically by this update.
FixedAn administrator could be silently demoted to manager, and the Back button on some pages left an administrator with the manager menu.
FixedInformation icons open when tapped, so they work on the tablets.
FixedThe per-station page reached from Relay Configuration was blank, and the header logo stayed on the page the user came from.
FixedOn Administrative Options, the character-type selectors for Define Work Order were drawn at the bottom of the page instead of on their row.
FixedThe message after a logo upload appeared in the red error box; it is now the green message that closes by itself.
Dispensing hardware (CDM)8
ImprovedThe controller re-synchronises the data stream from the dispensing hardware when it becomes garbled, instead of needlessly reconnecting. This resolves intermittent "checking hardware" hangs and failed pump starts.
NewA lost hardware link is detected and recovered automatically where it can be. An alert e-mail goes out if the hardware stays unreachable, and a "RESOLVED: dispensing hardware reconnected" e-mail when it comes back, with how long the outage lasted.
NewAn unplugged or disconnected CDM adapter is recognised. The controller does not power-cycle itself over it; after a short delay it e-mails LiquiDynamics, the brand and the site's Email Scheduler staff to check the cable and connection.
ChangedAfter a restart a "system starting up" notice shows until the controller is communicating with the dispensing hardware, so a dispense is not started too early. A new controller with no hardware configured yet is not held back by it.
ImprovedThe red hardware warning says what is actually wrong (not detected, just reconnected and still coming up, connected but not responding, or middleware not running) and gives one action to take, in the controller's language.
FixedDuplicate and contradictory hardware alerts around restarts and reconnects, and a "ready" before the hardware was really communicating again.
FixedReconnecting the CDM could power-cycle the controller, twice within a few minutes; brief interruptions left no record; an outage is now measured from when the connection was actually lost.
FixedAn interrupted read from the CDM could stall the controller for several seconds and mark every device offline, refusing a dispense that should have been allowed.
Updating the controller12
NewA rebuilt update system. The update page shows live progress with the download size, percent complete and a running clock, runs to completion even if the window is closed, cannot be started twice, and says "Update stopped -- nothing was installed" when it fails. It shows when the controller restarts and returns to the login page once it is back. Downloads that a network cuts part-way continue where they stopped, a stored update file is never deleted, and files copied onto the controller by hand no longer make an install fail part-way. The update pages appear in the controller's language; the release notes remain in English.
NewAfter an update the controller e-mails a confirmation to LiquiDynamics and your brand with a current database backup and its logs, once it is fully up; it names the version the controller came from and the one installed. On a network that blocks mail it stops trying after 24 hours and records one "Email sending FAILED" line in the Action Journal.
NewWhat changed, reached from the header of every screen, lists the release notes stored on the controller, so it works without internet access: a count of what is new, improved, changed and fixed, the day-one notices first, then one section per area, with a search, a filter and printing. A Release picker shows any earlier release, or a release and every one after it, and All Release Changes opens the notes of every release on oilcopsupport.com. After an update a What's New notice, in the controller's language, lists the changes, the ones that need attention first; after an update that skipped releases it covers every release since the one the controller was on. Each administrator and manager dismisses it once, after opening the notes. Technicians are not shown it.
NewInstall from a file, on Maintenance, Update System, for a site whose network blocks the update server: get the update zip from OilCop support or download it on a computer, then choose it or drag it onto the page. The controller checks that it is an OilCop update and which version, and installs nothing until you press Install now. An older version needs a warning and a confirmation, and is marked DOWNGRADE in the update e-mail.
NewWhen an update cannot be downloaded, the page tests the network and writes an e-mail for your IT department: the controller's IP address, MAC, gateway and DNS servers, exactly what failed, what to allow by name, and how IT can test it from their own computer.
ChangedThe update records the release line it installed, so the controller is offered the right releases afterwards.
ChangedAdministrator names, e-mail addresses and mail server settings are kept through updates.
ImprovedThe update switches on the web-server modules the controller's HTTPS protections need, then checks that a page still loads over HTTPS with those protections applied. If it does not, it puts the HTTPS setup back as it was, so every page keeps working, the update e-mail says so in its subject line, and the controller tries again at every restart until the protections can be applied.
FixedThe database update could be skipped silently during an install, leaving the new software running on the old database. The install now checks that the database update completed before it marks the version installed or restarts the controller; if not, the page says so in red with a link that runs it.
FixedAfter an update from an older release, the controller software could start before the database was ready and fail to come up. The start-up sequence is repaired on the first boot, and files and large logs left over from the old update method are cleaned up.
ImprovedA controller on any older release updates straight to this one. The database update runs a second time at the first start, after the start-up repair, so no change is skipped on an old database; before, a controller that jumped from an older R release kept 5-minute Dispense Page and Session timeouts instead of 10 and 20 minutes. The update e-mail reports the result.
ChangedThe update checks ask only for releases of the controller's own line, taken from the name of the installed release, so an update check can no longer move a controller to another line.
Database and backups6
ChangedDatabase backups are stored outside the web folder. Before this, a backup could be downloaded from the controller's web address without logging in.
ChangedFull backups are taken and restored with the database's own tools, which are faster and complete, and a restore can no longer destroy a database without saying so. The Restore Database list is sorted newest first, and a restore reports success or failure.
ImprovedPages and reports are faster on large databases, and the database update is more resilient on large or busy controllers; it records what it changed and what failed.
FixedThe controller software blocked its own database housekeeping at start-up, by locking a table it had just read, so one schema change was retried at every start without ever succeeding.
FixedThe controller software reconnects to the database instead of failing forever, and errors inside it are logged instead of vanishing.
FixedThe last four database tables that used a different storage engine from the rest have been moved onto the same one.
Reliability and security19
NewWhen this controller has been unable to reach its mail server, the reporting portal or the update server for more than an hour, administrators and managers see the orange notice described under Read first, saying what is blocked, since when, and what to ask your IT department to allow, with a ready-made e-mail to copy. It can be dismissed for 24 hours, 3 days or 7 days; a dismissal is in the Action Journal, applies to everyone on the controller, and hides only the problems shown at that moment. It clears itself when the connection is back. When the mail server can be reached but refuses the controller's login, the notice says who needs to act: your IT department for your own mail server, OilCop support for the OilCop mail service.
NewA "Controller clock" section on System Language & Time shows where the controller gets its time and the last automatic correction, and takes a Time server setting for networks that block the public ones (ask your IT department for an internal one; Windows domain controllers are time servers). "Set the time by hand", for a controller that reaches no time source, needs a PIN from OilCop support and refuses while oil is being dispensed. Every clock change of 30 seconds or more is written to the Action Journal, whoever made it, and a controller with no time source for a day shows a Clock line in the orange notice.
NewA background watchdog restarts the controller software if it stops. The watchdog itself keeps running and records a warning when it cannot, and an update can no longer start the controller software twice at once.
ChangedData reaches the reporting portal through a background transfer that sends only what changed, instead of the whole database. The old transfer, which carried a login in clear text, has been removed from every controller; the portal credential it held is carried over first. When the portal's copy lacks records the controller has already sent (a replacement controller that took over a site's database, a restore, records sent under an earlier database name, or a portal copy that was reset), those tables are sent again in full, once.
NewThe controller tells oilcopsupport.com which physical unit is sending its data (a code derived from its disk's serial number, nothing else), so OilCop support can see two controllers that share one network identity and would overwrite each other's data there. Nothing is refused; the transfer carries on.
ChangedDatabase and e-mail passwords are no longer kept in program files, and an old information page (test.php), an out-of-date image library (TimThumb) and unused scripts have been removed.
ChangedNetwork Settings. A saved change is applied at once, and the controller checks that the new address, mask, gateway and DNS are in effect, recording "applied and verified" or "NOT applied" with the reason in the Action Journal. A change that cannot be brought up is undone. If the controller must restart to complete it, it waits until no dispense is running and the top-off window has passed. An incomplete or invalid address, a gateway outside the subnet or 0.0.0.0 is refused on the page.
ChangedMaintenance actions that delete data require confirmation and a login.
ChangedThe "System is starting up" screen shows how long ago the controller started and that it keeps checking, so a slow start no longer looks like a hang.
ImprovedLog files are rotated and old ones removed automatically, memory use is tuned, and an always-on dispense activity log helps support review a dispensing problem.
FixedThe old address for setting the controller's time, the one that saves System Language & Time, and one internal support address accepted requests without a login; they now require the proper login or have been closed. An old address that showed the Email Scheduler's recipients to anyone on the network has been removed.
FixedSetting the controller's clock forward by ten minutes or more raised a false "hardware link DOWN" alert and restarted the controller.
FixedOn System Language & Time, Apply after changing the Daylight Saving switch showed "year must be numeric" and saved nothing.
FixedOn controllers upgraded from much older versions, commands sent from the web pages could not be cleared from their queue, so the same command could be carried out repeatedly. Such controllers are repaired by this update. A waiting command older than 60 seconds is now refused rather than carried out late; commands that stop a dispense are always carried out.
FixedA Network Settings change saved in the first minutes after a start was thrown away without a word; while the controller is starting, the page now saves nothing and says so.
FixedA controller with no dispensing devices configured yet reported "dispensing hardware disconnected" in the System Log and by e-mail.
FixedA mail failure reported as "Network is unreachable" was treated as the mail server refusing the message, so the notice gave the wrong advice.
FixedAn OilCop dialog could open hidden behind the orange or the red notice at the top of the page.
FixedThe file collecting the controller software's console output was never rotated, and a repeating fault could fill the disk; it now rolls over at 20 MB. The web application's log is no longer filled with one repeated notice.
Nothing matches these filters.