Raid 1, was ist los?

[BoBo_82]

Hallo,

nach mehreren Monaten uptime finde ich auf einmal folgendes in den logs:

Jun  1 01:06:01 webserver kernel: md: syncing RAID array md0
Jun  1 01:06:01 webserver kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
Jun  1 01:06:01 webserver kernel: md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reconstruction.
Jun  1 01:06:01 webserver kernel: md: using 128k window, over a total of 272960 blocks.
Jun  1 01:06:01 webserver kernel: md: delaying resync of md1 until md0 has finished resync (they share one or more physical units)
Jun  1 01:06:01 webserver kernel: md: delaying resync of md2 until md1 has finished resync (they share one or more physical units)
Jun  1 01:06:01 webserver kernel: md: delaying resync of md1 until md0 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md3 until md2 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md1 until md0 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md2 until md1 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md4 until md1 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md5 until md4 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md4 until md1 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md2 until md5 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md1 until md4 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md3 until md5 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md2 until md5 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md4 until md0 has finished resync (they share one or more physical units)
Jun  1 01:06:02 webserver kernel: md: delaying resync of md5 until md4 has finished resync (they share one or more physical units)
Jun  1 01:06:20 webserver kernel: md: md0: sync done.
Jun  1 01:06:20 webserver kernel: md: delaying resync of md5 until md4 has finished resync (they share one or more physical units)
Jun  1 01:06:20 webserver kernel: md: syncing RAID array md4
Jun  1 01:06:20 webserver kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
Jun  1 01:06:20 webserver kernel: md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reconstruction.
Jun  1 01:06:20 webserver kernel: md: using 128k window, over a total of 393472 blocks.
Jun  1 01:06:20 webserver kernel: md: delaying resync of md2 until md5 has finished resync (they share one or more physical units)
Jun  1 01:06:20 webserver kernel: md: delaying resync of md3 until md5 has finished resync (they share one or more physical units)
Jun  1 01:06:20 webserver kernel: md: delaying resync of md1 until md4 has finished resync (they share one or more physical units)
Jun  1 01:06:20 webserver kernel: RAID1 conf printout:
Jun  1 01:06:20 webserver kernel:  --- wd:2 rd:2
Jun  1 01:06:20 webserver kernel:  disk 0, wo:0, o:1, dev:hda1
Jun  1 01:06:20 webserver kernel:  disk 1, wo:0, o:1, dev:hdb1
Jun  1 01:06:44 webserver kernel: md: md4: sync done.
Jun  1 01:06:44 webserver kernel: md: syncing RAID array md1
Jun  1 01:06:44 webserver kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
Jun  1 01:06:44 webserver kernel: md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reconstruction.
Jun  1 01:06:44 webserver kernel: md: using 128k window, over a total of 4883648 blocks.
Jun  1 01:06:44 webserver kernel: md: delaying resync of md3 until md5 has finished resync (they share one or more physical units)
Jun  1 01:06:44 webserver kernel: md: delaying resync of md2 until md5 has finished resync (they share one or more physical units)
Jun  1 01:06:44 webserver kernel: md: delaying resync of md5 until md1 has finished resync (they share one or more physical units)
Jun  1 01:06:44 webserver kernel: RAID1 conf printout:
Jun  1 01:06:44 webserver kernel:  --- wd:2 rd:2
Jun  1 01:06:44 webserver kernel:  disk 0, wo:0, o:1, dev:hda8
Jun  1 01:06:44 webserver kernel:  disk 1, wo:0, o:1, dev:hdb8
Jun  1 01:11:37 webserver kernel: md: md1: sync done.
Jun  1 01:11:38 webserver kernel: md: syncing RAID array md5
Jun  1 01:11:38 webserver kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
Jun  1 01:11:38 webserver kernel: md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reconstruction.
Jun  1 01:11:38 webserver kernel: md: using 128k window, over a total of 68541184 blocks.
Jun  1 01:11:38 webserver kernel: md: delaying resync of md2 until md5 has finished resync (they share one or more physical units)
Jun  1 01:11:38 webserver kernel: md: delaying resync of md3 until md5 has finished resync (they share one or more physical units)
Jun  1 01:11:38 webserver kernel: RAID1 conf printout:
Jun  1 01:11:38 webserver kernel:  --- wd:2 rd:2
Jun  1 01:11:38 webserver kernel:  disk 0, wo:0, o:1, dev:hda5
Jun  1 01:11:38 webserver kernel:  disk 1, wo:0, o:1, dev:hdb5
Jun  1 01:39:33 webserver -- MARK --
Jun  1 01:59:34 webserver -- MARK --
Jun  1 02:19:34 webserver -- MARK --
Jun  1 02:24:48 webserver kernel: md: md5: sync done.
Jun  1 02:24:48 webserver kernel: md: delaying resync of md3 until md2 has finished resync (they share one or more physical units)
Jun  1 02:24:48 webserver kernel: md: syncing RAID array md2
Jun  1 02:24:48 webserver kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
Jun  1 02:24:48 webserver kernel: md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reconstruction.
Jun  1 02:24:48 webserver kernel: md: using 128k window, over a total of 2931712 blocks.
Jun  1 02:24:48 webserver kernel: RAID1 conf printout:
Jun  1 02:24:48 webserver kernel:  --- wd:2 rd:2
Jun  1 02:24:48 webserver kernel:  disk 0, wo:0, o:1, dev:hda9
Jun  1 02:24:48 webserver kernel:  disk 1, wo:0, o:1, dev:hdb9
Jun  1 02:27:51 webserver kernel: md: md2: sync done.
Jun  1 02:27:51 webserver kernel: md: syncing RAID array md3
Jun  1 02:27:51 webserver kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
Jun  1 02:27:51 webserver kernel: md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reconstruction.
Jun  1 02:27:51 webserver kernel: md: using 128k window, over a total of 1124416 blocks.
Jun  1 02:27:51 webserver kernel: RAID1 conf printout:
Jun  1 02:27:51 webserver kernel:  --- wd:2 rd:2
Jun  1 02:27:51 webserver kernel:  disk 0, wo:0, o:1, dev:hda6
Jun  1 02:27:51 webserver kernel:  disk 1, wo:0, o:1, dev:hdb6
Jun  1 02:28:58 webserver kernel: md: md3: sync done.
Jun  1 02:28:58 webserver kernel: RAID1 conf printout:
Jun  1 02:28:58 webserver kernel:  --- wd:2 rd:2
Jun  1 02:28:58 webserver kernel:  disk 0, wo:0, o:1, dev:hda7
Jun  1 02:28:58 webserver kernel:  disk 1, wo:0, o:1, dev:hdb7

Was hat das denn zu bedeuten? Wieso gibt es auf einmal ein resync des raids?
Das ganze läuft mit debian etch.

Für Tipps wäre ich dankbar.

[mattias1]

ein

cat /proc/mdstat

hilft weiter...

[BoBo_82]

Das hatte ich vergessen, alle raids funktionieren nach mdstat noch.

Personalities : [raid1] [raid6] [raid5] [raid4] [raid0]
md5 : active raid1 hda9[0] hdb9[1]
      68541184 blocks [2/2] [UU]

md4 : active raid1 hda8[0] hdb8[1]
      393472 blocks [2/2] [UU]

md3 : active raid1 hda7[0] hdb7[1]
      1124416 blocks [2/2] [UU]

md2 : active raid1 hda6[0] hdb6[1]
      2931712 blocks [2/2] [UU]

md1 : active raid1 hda5[0] hdb5[1]
      4883648 blocks [2/2] [UU]

md0 : active raid1 hda1[0] hdb1[1]
      272960 blocks [2/2] [UU]

unused devices: <none>

Ich finde es nur sehr merkwürdig, dass die raids auf einmal aus heiterem Himmel einen resync ausführen.

[Doc Brown]

Hi Leute,

ich habe genau das gleiche Problem. Mein Raid 1 hat auch um die exakt gleiche Uhrzeit neu syncronisiert.

cat /proc/mdstat sagt auch bei mir, daß angeblich alles erste Sahne ist.

alles ziemlich mysteriös, oder?

Hat da jemand ne Idee?????

Danke und Grüße
Doc

[shogun]

Ich hab das umgekehrte Problem:

laut mdstat ist die erste Platte ausgefallen, aber wenn ich die Platten teste, werden beide als OK angezeigt.

Hat da vielleicht jemand ne Idee?

[othan]

Bei mir läuft default mässig ein Cron job der jeden ersten Sonntag im Monat das Array überprüft

Schau mal nach /etc/cron.d/mdadm

/EDIT: @shogun: mal die erste platte aus dem array nehmen und wieder einhängen, wenns den resync nicht übersteht wechseln (oder gleich eine neue platte rein, je nachdem wie wichtig der server / die daten sind).

[shogun]

Hab durch Zufall nem Bekannten davon erzählt.
Der hat gemeint, ich solle mal ein raid rebuilt versuchen.
Ihm wäre das Problem bekannt.
Es trete immer dann auf, wenn man den Server einfach ausschaltet bzw. der Strom ausfällt.

Also kurze Web-Recherche gefahren,

mdadm <Raiddevice> --add <Partition>

ausgeführt, und siehe da, er recovered die Partitionen der Platte und sie wird wieder als aktiv angezeigt.