Support for mirrored swap partitions
Hey, so I opened a thread about this issue: https://forums.freenas.org/index.php?threads/9-10-system-crash-with-unexplained-log.53865/#post-372899
The server crashed last night, destroying file transfers and killing jails, and also the GUI was unresponsive - ping were acknowledged, mails kept getting sent.
I'm openning this bug to attach the debug info, and also to get this issue to your attention guys because its really severe.
After reseting the unit I got the errors on the on-screen console, picture is attached.
feat(middlewared): private method to create mirror for swaps
feat(middlewared): auxiliary method to remove disk from swap
fix(notifier): use middleware to detach swap
While I'm here remove most of the UFS volumes codepath.
fix(rc.d): remove now unused function and use ld path for jq
feat(middlewared): set up geli and online swap
While I'm here fix bug of name of mirror on existing swaps and skip
feat(middlewared): use single disks if swap mirror could not be formed
feat(middlewared): change disk.swaps_remove_disks to accept a list
fix(middlewared): make sure socket file exists before trying
fix(middlewared): wait socket to be ready for read
Make this gevent friendly or it can block indefinitely.
fix(middlewared): configure dumpdev on the first disk of gmirror
Or in the first unused partition.
fix(middlewared): remove degraded swap mirror and start again
#3 Updated by Alexander Motin over 1 year ago
- Tracker changed from Bug to Feature
- Subject changed from FN Crashed completely with strange log to FreeNAS needs mirrored log partitions
- Category changed from 53 to 201
- Status changed from Unscreened to Screened
- Priority changed from No priority to Nice to have
- Target version set to 11.1
Yes. Looking on provided logs and messages it seems like you lost one of your SATA disks. Because that disk also included swap partition that was used for some reason, some system daemons went south. We obviously should do something to increase reliability of swap partitions, but at this point system did what it supposed to, so it is more misfeature then a bug.