[PVE-User] Again trouble, but this time with ext4/trim...

Marco Gaiarin gaio at sv.lnf.it
Mon Jul 1 11:00:48 CEST 2019


Trying to fix my trouble with Win2016, last week i've reboot completely
my PVE 5 cluster.

Cluster use a SAN (HP MSA 1040), thin storage mode, used as LVM in PVE,
eg:
  lvm: VM0
	vgname VM0
	content images,rootdir
	shared 1

Because the underlying storage are thin, i've do some test in the past
and enable 'discard' on PVE volumes, verifying that storage space get
effectively freed.


After that reboot, at every execution of 'fstrim' (via systemd timer)
in two different VMs i got:

 Jul  1 00:00:03 vdmsv2 kernel: [542165.662171] EXT4-fs error (device sda6): ext4_mb_generate_buddy:759: group 647, block bitmap and bg descriptor inconsistent: 21279 vs 32768 free clusters
 Jul  1 00:00:03 vdmsv2 kernel: [542165.695216] EXT4-fs error (device sda6): ext4_mb_generate_buddy:759: group 648, block bitmap and bg descriptor inconsistent: 12011 vs 32768 free clusters
 Jul  1 00:00:03 vdmsv2 kernel: [542165.705122] EXT4-fs error (device sda6): ext4_mb_generate_buddy:759: group 649, block bitmap and bg descriptor inconsistent: 14971 vs 32768 free clusters
 Jul  1 00:00:03 vdmsv2 kernel: [542165.711123] EXT4-fs error (device sda6): ext4_mb_generate_buddy:759: group 650, block bitmap and bg descriptor inconsistent: 12194 vs 32768 free clusters
 Jul  1 00:00:06 vdmsv2 kernel: [542169.243447] JBD2: Spotted dirty metadata buffer (dev = sda6, blocknr = 0). There's a risk of filesystem corruption in case of system crash.
 Jul  1 00:00:06 vdmsv2 kernel: [542169.244995] JBD2: Spotted dirty metadata buffer (dev = sda6, blocknr = 0). There's a risk of filesystem corruption in case of system crash.

(sda6 is /var)

 Jul  1 00:01:05 vdmsv1 fstrim[3806]: fstrim: /var/spool/squid: FITRIM ioctl failed: Messaggio errato
 Jul  1 00:01:05 vdmsv1 kernel: [9519779.612188] EXT4-fs error (device sdb1): ext4_validate_block_bitmap:386: comm fstrim: bg 506: bad block bitmap checksum
 Jul  1 00:01:05 vdmsv1 kernel: [9519779.636314] EXT4-fs warning (device sdb1): ext4_trim_all_free:5130: Error -74 loading buddy information for 506
 Jul  1 00:01:08 vdmsv1 kernel: [9519782.441786] EXT4-fs error (device sda6): ext4_mb_generate_buddy:759: group 181, block bitmap and bg descriptor inconsistent: 32767 vs 32768 free clusters
 Jul  1 00:01:08 vdmsv1 kernel: [9519782.448571] JBD2: Spotted dirty metadata buffer (dev = sda6, blocknr = 0). There's a risk of filesystem corruption in case of system crash.
 Jul  1 00:01:09 vdmsv1 kernel: [9519782.973496] EXT4-fs error (device sda6): ext4_mb_generate_buddy:759: group 278, block bitmap and bg descriptor inconsistent: 32749 vs 32768 free clusters
 Jul  1 00:01:09 vdmsv1 kernel: [9519783.034358] EXT4-fs error (device sda6): ext4_mb_generate_buddy:759: group 325, block bitmap and bg descriptor inconsistent: 32740 vs 32768 free clusters
 Jul  1 00:01:09 vdmsv1 kernel: [9519783.238781] JBD2: Spotted dirty metadata buffer (dev = sda6, blocknr = 0). There's a risk of filesystem corruption in case of system crash.

(sdb1 is /var/spool/squid and sda6 is /var)


In that servers i've also some other FS, but the ext4 ones are low
varying, mounted RO or noatime, but also some FS in XFS fs, that seems
does not suffer.


Every time i run 'fstrim -av' by hand i've never catch an error, so
seems to me that trouble happen whene the system are under high IO (as
when run many weekly tasks).


For now, i disable fstrim timer. But seeking some help...

-- 
dott. Marco Gaiarin				        GNUPG Key ID: 240A3D66
  Associazione ``La Nostra Famiglia''          http://www.lanostrafamiglia.it/
  Polo FVG   -   Via della Bontà, 7 - 33078   -   San Vito al Tagliamento (PN)
  marco.gaiarin(at)lanostrafamiglia.it   t +39-0434-842711   f +39-0434-842797

		Dona il 5 PER MILLE a LA NOSTRA FAMIGLIA!
      http://www.lanostrafamiglia.it/index.php/it/sostienici/5x1000
	(cf 00307430132, categoria ONLUS oppure RICERCA SANITARIA)


More information about the pve-user mailing list