- OvmsDev - Open Vehicles

Automatic naming of CAN log files
by Ludovic LANGE 03 Dec '22

03 Dec '22

Hi, I'm planning to implement an "automatic naming" of the CAN log files for the VFS target. It would also take care of the file name rotation according to some criterion (file size, and/or log duration). I've created an issue to follow it here https://github.com/openvehicles/Open-Vehicle-Monitoring-System-3/issues/748 (and some others also) but let me first share what I have in mind: The use case is the following: I want to setup OVMS as a CAN bus datalogger. It should be always-on, always and continuously logging, event after a restart. (For that, my current solution is to use event scripts, cf https://github.com/openvehicles/Open-Vehicle-Monitoring-System-3/issues/766 ) Logging to the same file, in that case, present some challenges: * file size would grow indefinitely, which could be then make it difficult to copy / network send somewhere * file would always be opened, which could prevent any action on it (if we want to send it somewhere else, or remove it) * in case of crash / corruption the whole file could be lost * in case of reboot you'd end overwriting the file. Of course we should change the file name, but it's not easy to do it in a script. It may be possible to do this with Duktape and `|OvmsCommand.Exec|` but I fear that we may miss events if we stop / start logging on a highly loaded bus. A proposal to solve this would be to add a new mode in `canlog_vfs.cpp` that would cater for this automatic naming (inspiration for this feature coming from another product). In this mode you don't pass the path as a parameter, but instead a `log prefix` that will be used to construct the file path. Syntax could be something like : `can log start vfs-auto crtd <log prefix> [filter1] ... [filterN]` (no filename). It would automatically generate a log file name, taking into account the splits / rotations that can be configured: * if asked for by configuration - rotate file name (= increment file name) after a file size is reached * if asked for by configuration - rotate file name (= increment file name) after a certain logging duration has been reached Naming would be 'automatic' with something like : `/sd/log/<vehicleid>/<#power on>/<log prefix>/<#rotations>.<log extension>` where: * `/sd/log/` is arbitrary. Could be coming from configuration if wanted (esp. for module without sd card) * `<vehicleid>` is there to help to declutter your sdcard (if you share it with multiple modules), or to declutter your storage server (when https://github.com/openvehicles/Open-Vehicle-Monitoring-System-3/issues/749 has been completed). We may have to make it a valid filename. * `<#power on>` is the number of time the module has powered on ; it is an always incrementing number expressed as an integer on 8 digits. Depending on the configuration, some modules will be always-on ; others will stop more frequently - the number will overflow after 99.999.999 power on, which will certainly see a change of sd card, so no fear of unwanted overwriting of files * `<log prefix>` is a way to distinguish between multiple concurrent can logs active. * `<#rotations>` is a counter of rotations for this file and this power on ; it is an integer on 8 digits ; reset to `00000001` at each poweroff and that will increment each time there is a file name rotation. some configuration items cater for the file name rotation: * `split_size_mb` : when the file has reached this size, the file is closed, the element `<#rotations>` is incremented, and a new file with the 'automatic' name is opened. * `split_duration_s` : when the file has logged for this duration, the file is closed, the element `<#rotations>` is incremented, and a new file with the 'automatic' name is opened. example of file name: `/sd/log/mycar/00000024/can1-25k/00000005.crtd` (Of course we need a way to either stop logging once the VFS is full, or to auto-remove some log files under this path) Let me know if you have any thoughts / concerns about this. Regards, Ludovic

2 5

FYI: Historical Messages on api.openvehicles.com
by Mark Webb-Johnson 28 Nov '22

28 Nov '22

FYI We had an outage last night due to issues with an excessively ovms_historicalmessages table on the api.openvehicles.com <http://api.openvehicles.com/> server. Currently, that table has more than 150 million records, and has become largely unworkable due to its size. I have temporarily moved the historical records to another table, and resumed service. Now working to find out the root cause of the issue (likely a few vehicles abusing the system). Hopefully I will be able to start restoring the old historical data later today. Regards, Mark.

1 0

Fasttech: End of an era
by Mark Webb-Johnson 24 Nov '22

24 Nov '22

Just a short note to say that, effective immediately, OVMS will no longer be available from Fasttech. They have provided incredible support for the project over the past 10+ years, and we are immensely grateful for all that they have done. The team there have been fantastic, gone above and beyond, and I highly recommend them. While a few issues have led to the decision not to sell through Fasttech any more, primarily (with COVID-19 driving international logistics costs and shipment delays up), supporting OVMS modules sold globally through Fasttech has become prohibitively expensive and simply painful for everyone concerned (us, Fasttech, and the buyer). We are still a hobbyist project, without paid employees to handle support. I would far rather that users in Europe buy through OpenEnergyMonitor, and users in North America buy through MedlockAndSons - obtaining quick simple delivery and local support. The vast majority of our users are in EU and US. We will be looking for some other way of providing for and supporting Asia Pacific customers (probably out of Hong Kong, to keep warehousing and distribution as simple as possible). I will let you know once a decision there has been made, and if anyone has any suggestions please let me know. Regards, Mark.

3 4

Kia Soul - Incorrect/Inconsistent Charging State (Kia Soul EV 30KWh)
by Mark Webb-Johnson 23 Nov '22

23 Nov '22

Anybody able to help out with this one? https://www.openvehicles.com/node/3443 Incorrect/Inconsistent Charging State (Kia Soul EV 30KWh) Thanks, Mark

1 0

Fiat500e Quirks
by William Reading 19 Nov '22

19 Nov '22

Howdy there, I recently got an OVMS set up with my Fiat 500e and was happy to see that it works, but I noticed noticed a few oddities. The first one is that it uses Valet Mode (CommandActivateValet) for preconditioning instead of CommandClimateControl: https://github.com/openvehicles/Open-Vehicle-Monitoring-System-3/blob/bd192… <https://github.com/openvehicles/Open-Vehicle-Monitoring-System-3/blob/bd192…> Is there a downside to changing the code to where it uses Climate Control instead? It seems like that would fit better with the intended functionality. The other is that it seems to rapidly generate spurious charging / not charging events, which en masse seems to make the websocket connection on the web server time out. I can put together a pull request to try to fix these issues, but was wondering if there are any thoughts on these two things before getting started. There's also no pictures of the Fiat on the iOS App, but I assume that's because the last release of the iOS App predates support. Thanks, William Reading

2 3

Strange crashes when logging additional info in can logs
by Ludovic LANGE 19 Nov '22

19 Nov '22

Dear list, You may remember that I'm trying to enhance the "datalogger" side of OVMS - in different ways. One facet of this endeavour is to have (CAN) logs that also incorporate other data coming from OVMS. In particular, I'd like to have GNSS (GPS) data in the same log as my CAN data. My first try is to slightly alter `canlog.cpp` and add the ability to `LogInfo` some chosen metrics, in the same spirit of the already logged subset of events. (With CRTD format, it appears as comments) While doing this, I'm facing multiple crashes that are somehow strange: * The crashes only occur when closing the logs (`can log stop`), never when logging * The cause of the crash is not always the same (LoadProhibited, InstrFetchProhibited) * Sometimes the PC is impacted / overwritten The crashes seems to always occur in a window of 20s after the `can log stop` command (In general it's immediate, but sometime can take some more seconds). (It lead me also to discover that the feature of our version of esp-idf that resolves the memory addresses in the serial output - `make monitor` - is broken, and that it's better to use GDB for that) While I am doing my homework and try to solve this, I wanted to share it here also to see if you already encountered this and/or if my changes are not done properly / not on the right OS task / etc.... - for which I lack background and experience. (In the process I think I found a non-freed variable, but it's not causing my issue - cf https://github.com/openvehicles/Open-Vehicle-Monitoring-System-3/pull/774 ) Here are the changes - you'll see that the crash do occur as soon as I `LogInfo` something - after the log has been closed. ------------------------------------------------------------------------ modified: vehicle/OVMS.V3/components/can/src/canlog.cpp ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── @ vehicle/OVMS.V3/components/can/src/canlog.cpp:355 @ canlog::canlog(const char* type, std::string format, canformat::canformat_serve_ using std::placeholders::_1; using std::placeholders::_2; MyEvents.RegisterEvent(IDTAG, "*", std::bind(&canlog::EventListener, this, _1, _2)); + MyMetrics.RegisterListener(IDTAG.c_str(), "*", std::bind(&canlog::MetricListener, this, _1)); int queuesize = MyConfig.GetParamValueInt("can", "log.queuesize",100); m_queue = xQueueCreate(queuesize, sizeof(CAN_log_message_t)); @ vehicle/OVMS.V3/components/can/src/canlog.cpp:365 @ canlog::canlog(const char* type, std::string format, canformat::canformat_serve_ canlog::~canlog() { MyEvents.DeregisterEvent(IDTAG); + MyMetrics.DeregisterListener(IDTAG.c_str()); if (m_task) { @ vehicle/OVMS.V3/components/can/src/canlog.cpp:441 @ void canlog::EventListener(std::string event, void* data) LogInfo(NULL, CAN_LogInfo_Event, event.c_str()); } +void canlog::MetricListener(OvmsMetric* metric) { + std::string name = metric->m_name; + ESP_LOGI(TAG,"MetricListener - enter : %s", name.c_str()); + if (startsWith(name, "v.p.gps") || (name == "v.p.latitude") || (name == "v.p.longitude") || (name == "v.p.altitude") || (name == "v.p.direction") || (name == "v.p.satcount") || (name == "m.net.wifi.sq") || (name == "m.freeram")) { + std::string metric_text = "metric: " + name; + ESP_LOGI(TAG,"MetricListener - keep : %s", metric_text.c_str()); + LogInfo(NULL, CAN_LogInfo_Event, metric_text.c_str()); + } +} + const char* canlog::GetType() { return m_type; Note: "m.net.wifi.sq" / "m.freeram" are providing debugging data, so that I can quickly receive metrics after the boot. Not what I'm interested in for the final version of course, and the list will be configurable. ------------------------------------------------------------------------ Here are some example of panic handler infos: Guru Meditation Error: Core 1 panic'ed (LoadProhibited). Exception was unhandled. Core 1 register dump: PC : 0x400d95f8 PS : 0x00060830 A0 : 0x8030d32c A1 : 0x3ffc4980 A2 : 0x3ffeb5a8 A3 : 0x00000000 A4 : 0xfefefefe A5 : 0x3ffb5200 A6 : 0x3f805e78 A7 : 0x000000fe A8 : 0x800d95f0 A9 : 0x3ffc4930 A10 : 0x3ffeb5a8 A11 : 0x00000000 A12 : 0x00000009 A13 : 0x73c80b46 A14 : 0x3ffc4930 A15 : 0x3f8b4164 SAR : 0x00000004 EXCCAUSE: 0x0000001c EXCVADDR: 0xfefeff2e LBEG : 0x4008abe4 LEND : 0x4008ac00 LCOUNT : 0xffffffff ELF file SHA256: eb3e5ce8f67640c8 Backtrace: 0x400d95f8:0x3ffc4980 0x4030d329:0x3ffc49e0 0x4012122d:0x3ffc4a00 0x40121296:0x3ffc4a60 0x40120e6e:0x3ffc4a80 0x400ede1f:0x3ffc4aa0 0x400eed41:0x3ffc4ae0 0x400ecdc9:0x3ffc4b00 0x4011c5b6:0x3ffc4b40 0x4011c7b6:0x3ffc4b70 0x4011c8d8:0x3ffc4be0 0x0 0x400d95f8 is exactly the address of the call to LogInfo in my changes: (gdb) list *0x400d95f8 0x400d95f8 is in canlog::MetricListener(OvmsMetric*) (vehicle/OVMS.V3/components/can/src/canlog.cpp:447). 442 std::string name = metric->m_name; 443 ESP_LOGI(TAG,"MetricListener - enter : %s", name.c_str()); 444 if (startsWith(name, "v.p.gps") || (name == "v.p.latitude") || (name == "v.p.longitude") || (name == "v.p.altitude") || (name == "v.p.direction") || (name == "v.p.satcount") || (name == "m.net.wifi.sq") || (name == "m.freeram")) { 445 std::string metric_text = "metric: " + name; 446 ESP_LOGI(TAG,"MetricListener - keep : %s", metric_text.c_str()); *447 LogInfo(NULL, CAN_LogInfo_Event, metric_text.c_str());* 448 } 449 } 450 0xfefe / 0xfefefefe seems related to some of the Heap Memory Debugging options I've activated (Cf https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-reference/s… ) and seems to indicate a possible use-after-free issue. I'll look to debug in that direction. Another version (before Heap Memory Debugging activated) did that from time to time: ------------------------------------------------------------------------ Guru Meditation Error: Core 0 panic'ed (InstrFetchProhibited). Exception was unhandled. Core 0 register dump: PC : 0x00008100 PS : 0x00060530 A0 : 0x802f0cf0 A1 : 0x3ffd0f00 A2 : 0x3ffeb214 A3 : 0x00008100 A4 : 0x402f1df8 A5 : 0x00000000 A6 : 0x00000003 A7 : 0x3ffb556c A8 : 0x800d9540 A9 : 0x3ffd0ee0 A10 : 0x3ffeb214 A11 : 0x00000000 A12 : 0x00000009 A13 : 0x3f8b4a24 A14 : 0xffffffff A15 : 0x40404040 SAR : 0x00000018 EXCCAUSE: 0x00000014 EXCVADDR: 0x00008100 LBEG : 0x4008a848 LEND : 0x4008a864 LCOUNT : 0xffffffff ELF file SHA256: f4d0c43aa48ef759 Backtrace: 0x00008100:0x3ffd0f00 0x402f0ced:0x3ffd0fb0 0x4011c56d:0x3ffd0fd0 0x4011c5c6:0x3ffd1020 0x4011c24a:0x3ffd1040 0x4014f51d:0x3ffd1060 0x4014f953:0x3ffd1080 0x4014e6e3:0x3ffd11d0 0x4014ef16:0x3ffd1210 0x4014f001:0x3ffd1230 0x4014f119:0x3ffd1260 0x4014e889:0x3ffd1290 0x4014e9ed:0x3ffd1310 0x4014ebf9:0x3ffd13f0 ------------------------------------------------------------------------ Thanks for any hint or advices you may want to share. Regards,

2 3

Fwd: Tesla Model 3 LR development
by Michael Balzer 13 Nov '22

13 Nov '22

Adam Pisl offers to help in developing the Model 3 adapter. Is anyone working on this or able to help Adam in getting an initial setup? If so please contact him directly (cc me if you like). Regards, Michael -------- Weitergeleitete Nachricht -------- Datum: Sun, 13 Nov 2022 08:27:00 +0100 (CET) Von: carprofi(a)carprofi.cz An: dexter(a)dexters-web.de Hello, I have purchased Tesla 3 LR, and want to participate on testing of OVMS development. Could You help me with getting the proper OVMS kit? Adam Pisl <carprofi(a)carprofi.cz> -- Michael Balzer * Helkenberger Weg 9 * D-58256 Ennepetal Fon 02333 / 833 5735 * Handy 0176 / 206 989 26

1 0

Advice regarding BmsSetCell...
by Mark Webb-Johnson 12 Nov '22

12 Nov '22

The original Model S had a fairly simple to decode CAN bus message 0x6f2 with the first byte in range 0..1f and each message encoding 3 battery bricks, for a total of 96 bricks. However, things have now got more complicated. A 75kWh Model X, and a replacement pack for a 85kWh Model S, contain cut-down packs with modules missing, so 0x6f2 only contains 28 messages, for 84 bricks, with bricks #15, #16, #17 and #31 missing from the message. When we feed this into BmsSetCell… calls, the OVMS BMS support code sees the missing bricks, assumes things are missing, and refuses to give an output. So my (hopefully simple) question, is how to tell it the brick arrangement? Or how to deal with these missing bricks? Regards, Mark.

3 3

Re: [Ovmsdev] Automatic naming of CAN log files
by Ludovic LANGE 10 Nov '22

10 Nov '22

I do agree with you, and I do the same on any system which has a battery-backed RTC and a persistent network connexion (taking care also of timezone issues)... However, I'm afraid that after a module reboot it may take a few seconds to retrieve a correct time. Also, when running without GPS, without network and without modem (which seems a supported, even if unusual, configuration), it seems that I'll never have a correct time at all. In both cases, the module will always start with Unix epoch (1970-01-01-00:00:00) for at least a few seconds, and so risk the risk to overwrite log files do exist (as the log starts as soon as possible). Of course, I'm running in that configuration (no modem, very often without Wifi, GPS not always having time) otherwise it would not be fun... :-( So that's why I wanted to be independent of time. But it would not be complex to support both syntaxes, to offer this choice. Ludovic Le 09/11/2022 à 22:57, Craig Leres a écrit : > On 11/9/22 13:52, Ludovic LANGE wrote: >> Naming would be 'automatic' with something like : >> `/sd/log/<vehicleid>/<#power on>/<log prefix>/<#rotations>.<log >> extension>` > > When I auto-name log or trace files I like to use the date and time. > This format sorts nicely and makes it easy to find the file that > contains the date+time of interest. So I would counter-propose: > > YYYY-MM-DD-HH:MM:SS.<log extension> > > Craig

1 0

HUD Connected to OVMS not working
by Michael Geddes 09 Nov '22

09 Nov '22

Hi all, Anybody got any experience with HUDs? I bought this simple HUD from FastTech: https://www.fasttech.com/products/9686283 When connected direct to the OBDII port, it powers on and seems to function. When I connect it to the OVMS module with the standard https://www.fasttech.com/products/9652027 the HUD doesn't even power on. It's stated draw is 200ma which afaict should be fine. I'm pretty sure the 12V power is getting to PIN 16... though I will try again to make sure. I only have a crappy little Multimeter. On the Troubleshooting it states: "HUD only available for cars with OBD2 and EU-OBD agreements ... HUD Products don't support JOBD and OBDI protocol." The OBD HUD connector has the following pins: 4* Chassis 5* Signal Ground 6 CAN High 7 *ISO9141 K Line* 14 CAN Low 15 *ISO9141 L Line* 16 Vehicle Battery +ve * Long pins These are the standard OVMS supplied pins: 4 Chassis Ground 5 Chassis Ground 6 CAN H (Primary) 14 CAN L (Primary) 16 +12V So obviously the main difference is the lack of K / L lines in the OVMS ... but I wouldn't have thought that would have prevented it from powering up. Does anybody have any thoughts on this? //.ichael

4 12