add additional stats from mdstat #380
Conversation
|
We need to make sure this works for all kernels we support. If that's the case, LGTM! |
|
This should gracefully handle different kernels...and I believe the mdstat file format hasn't changed in quite some time. |
Counters added: * excessive_buffer_overrun_errors * local_link_integrity_errors Signed-off-by: Trey Dockendorf <[email protected]> Signed-off-by: John Seekins <[email protected]>
Signed-off-by: John Seekins <[email protected]>
Signed-off-by: John Seekins <[email protected]>
Signed-off-by: John Seekins <[email protected]>
|
@discordianfish Is there a standard way to prove out the "kernels supported"? I know this works on 4.19, for example. |
|
Typically it involves a lot of browsing the kernel source tree. :-( To note, we want to keep kernel support all the way back to 2.6.23. |
|
Unfortunately I don't think we have a way. We should probably collect fixtures for multiple kernel versions. It's silly but I'd probably spin up a VM with e.g 2.6.23 and run tests there.. |
|
If that's the case...how did y'all validate this collector in the first place? While I appreciate that we should try to validate this as much as possible, it does fail gracefully on missing stats. |
|
Lots of trial and error. The current mdstat fixture is a collection of various examples from bugs reported by users. |
There was a problem hiding this comment.
Fair enough, I think we should extend the fixtures to cover more kernel versions but until then, I think it's fair to support the procfs files as shown in the fixtures. So LGTM.
@SuperQ wdyt?
* Add several Infiniband counters Counters added: * excessive_buffer_overrun_errors * local_link_integrity_errors Signed-off-by: Trey Dockendorf <[email protected]> Signed-off-by: John Seekins <[email protected]> * add additional stats from mdstat Signed-off-by: John Seekins <[email protected]> * return successful values every time Signed-off-by: John Seekins <[email protected]> * add count of 'downed' disks Signed-off-by: John Seekins <[email protected]> Co-authored-by: Trey Dockendorf <[email protected]>
* Add several Infiniband counters Counters added: * excessive_buffer_overrun_errors * local_link_integrity_errors Signed-off-by: Trey Dockendorf <[email protected]> Signed-off-by: John Seekins <[email protected]> * add additional stats from mdstat Signed-off-by: John Seekins <[email protected]> * return successful values every time Signed-off-by: John Seekins <[email protected]> * add count of 'downed' disks Signed-off-by: John Seekins <[email protected]> Co-authored-by: Trey Dockendorf <[email protected]>
In the recovery line for mdstat, we can also track percentage complete, estimated time to completion, and current recovery write speed. This MR adds those additional stats.