Change sai log level before discovery to prevent SAI ERR spam to the log#316
Conversation
|
@kcudnik to check. |
|
@pavel-shirshov @lguohan This pull request will reset all the SAI components' loglevel back to NOTICE. It breaks loglevel for warm-reboot Suggest store the previous loglevel, and restore back. |
| { | ||
| SWSS_LOG_TIMER("discover"); | ||
|
|
||
| set_sai_api_log_min_prio("SAI_LOG_LEVEL_CRITICAL"); |
There was a problem hiding this comment.
because otherwise sai will dump to syslog many thousands of the error messages "something not found" when syncd discover sai api". Output to syslog seriously hits cpu which increases fast-reboot time
There was a problem hiding this comment.
there is nothing that produces critical message, so nothing would be logged except maybe something from internal sai
i would propose to log SAI_STATUS_NOT_IMPLEMENTED (which is the most errors probably) as info message in discovery method, instead of disabling all messages in this way
There was a problem hiding this comment.
Hi Kamil,
Yes. I made that level CRITICAL because no messages are produced by SAI on syncd discovery.
I had a patch making SAI_STATUS_NOT_IMPLEMENTED as INFO in SAI. I'm not sure it's still there.
There was a problem hiding this comment.
yea that would be better approach, so other meaningfull errors if present they still would be logged
4 participants
Change the level of sai api log to critical right before discovery and set it back to notice after the discovery to prevent error messages caused by discovery process