[smartmontools-support] Dealing with locked raid device files
vindrg at gmail.com
Mon Dec 31 13:15:46 CET 2018
I have reported this Debian wishlist bug report:
Is it possible, and at all reasonable, to deal with these kind of niche problems?
In short, Areca's `cli64` command line interface, that queries hardware raid controller information
(raidset, volume, hardware status, etc), locks `/dev/sgX` device with exclusive lock, and that
prevents `smartctl` and `smartd` to query device, and in the result, errors happens - missing
monitoring checks (for munin and similar), and email "spam" from smartd with "SMART error
(FailedOpenDevice) detected on host..." messages.
In our setup, `cli64` is invoked every minute to catch raidset errors, and send SMS if so, as fast
as possible via Icinga monitoring system. In the result, there are *two* `cli64` invocations per
minute, to get Areca log AND raidset status (these provide different set of important messages).
Maybe it could be possible to detect exclusive locks and retry/wait (for a reasonable
attepts/duration) to reduce "false-possitives"? Or this would be considered as "dirty" workaround..?
More information about the Smartmontools-support