#917 closed patch (fixed)
20s timeout is too short for large JBODs
| Reported by: | asomers | Owned by: | |
|---|---|---|---|
| Priority: | minor | Milestone: | Release 6.6 |
| Component: | all | Version: | 6.5 |
| Keywords: | scsi | Cc: |
Description
Most smartctl commands have a default timeout of 20 seconds. This is sufficiently long for pretty much any disk to spin up from standby. However, many large JBODs implement staggered spinup, where they limit the number of drives that can simultaneously spin up. On such systems, it may take considerably longer for all drives to spin up. For example, I have a 96-drive JBOD where spinning up all drives takes nearly 45 seconds. The attached patch fixes smartctl for me.
Attachments (1)
Change History (4)
Changed 6 years ago by
| Attachment: | patch-scsicmds.h added |
|---|
comment:1 Changed 6 years ago by
I think it is fine. Moreover, in some code controller-specific this value is not respected and 60s is hardcoded. I will commit it
comment:2 Changed 6 years ago by
| Component: | smartctl → all |
|---|---|
| Keywords: | scsi added |
| Milestone: | → Release 6.6 |
| Resolution: | → fixed |
| Status: | new → closed |
Thank you for the patch, fixed in r4557
Note: See
TracTickets for help on using
tickets.
Increase default timeout from 20s to 60s