Extend API to change the SysLog Log Level

[proddy]

I propose a new system command called syslog_level called via /ems-esp/api/system/syslog_level and take a JSON body as

{ "log": "<level str>" }

"log" can also be "value" to be compatible with the other calls.

where <level str> is one of OFF, EMERG, ALERT, CRIT, ERR, WARNING, NOTICE, INFO, DEBUG, TRACE, ALL.

[proddy]

Perhaps also a special called RAW which does the same as watch raw in the console. i.e. only the telegrams as raw bytes are sent to the SysLog.

[MichaelDvP]

The raw option affects all loggers (console/web/syslog) and also the watch command (which affects also all loggers). I think it's confusing to have this as extra level in syslog command. Maybe better to have the raw-switch as extra api command (/api/system/log_raw: {"value":true}) or (imo not so good) as extra parameter: /api/system/syslog: {"level":"all","raw":true}
We can also add a parameter to syslog to transmit only the selected level, not all levels below: {"level":"trace", "raw":true", "filter":true} (as a hack we should allow for filtered-trace also the notice logs if watch is activated)

[proddy]

Ok. so for now

1. add /api/system/log_raw: {"value":true}
2. add /api/system/syslog_level: {"log": "<level str>"}

that will satisfy what Thomas needs if he uses NOTICE and turns on raw. The filter is a nice idea but I don't think it's worth the extra work for his edge case. Unless others will use it.

[MichaelDvP]

if he uses NOTICE and turns on raw

Ups, we have a misunderstanding, i thought about the "raw" option in web-settings, that only switches "watch" from pretty to hex, but do not activate watch. (btw: I've used raw a long time, but pretty_telegram is stable and trustworthy since a long time now, raw is mostly obsolete).
For this it's better to change 1. to api/system/watch: {"value": "off|on|raw"} to stay in sync with console. Than log notice and watch on will work like in console and affects console, web-log and syslog.

[proddy]

ok. agree. keep it in sync is smart

[tp1de]

@proddy @MichaelDvP Are any changes to syslog done in v3.2.2a0 ?

I was testing for km200 delay in #90 and recognized that syslog output changed.

Syslog level trace and output in hex works fine as I proposed - just the rx messages - FINE !

The other loglevels send wrong format to syslog server so I get error messages from syslog.
In my case node-red syslog node with log-level set to debug for example:

"Invalid message received: {"timestamp":"2021-08-21T19:28:06.175Z","address":"192.168.178.71","family":"IPv4","port":59707,"size":136,"msg":"<7>1 2021-08-21T21:27:00.026586+2:00 ems-esp emsesp - - - 000+00:35:32.043 D 8353 No telegram type handler found for ID 0x8E4 (src 0x08)"}"

Isn't the following missing?: facility: "logalert",facilityCode: 14,severity: "debug",severityCode: 7,tag: "Rx"

[MichaelDvP]

My alpha-version sometimes differ from official version. This #90 test was intended only to test the km200 handling, please switch back to offiicial beta for syslog discussions.

[tp1de]

@proddy @MichaelDvP

Syslog errors (wrong format) are also related to last dev version v3.2.2b0.

[proddy]

@tp1de can this be closed?

[tp1de]

Yes