Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
> -----Message d'origine----- > De : drbd-user-bounces at linbit.com > [mailto:drbd-user-bounces at linbit.com] De la part de Lars Ellenberg > > On Mon, Dec 10, 2007 at 08:43:32AM -0500, Fortier,Vincent > [Montreal] wrote: > > It seems to happend only on the Primary node. It does > > happend on both 2.6.21.7 & 2.6.22.14 kernels. > > > > Here is current dmesg: > > [410696.551852] drbd1: drbd_bm_set_bits_in_irq: (!b->bm) in > > drivers/block/drbd/drbd_bitmap.c:1028 > > [410696.551901] drbd1: drbd_bm_set_bits_in_irq: (!b->bm) in > > drivers/block/drbd/drbd_bitmap.c:1028 > > [410696.553440] drbd1: drbd_bm_set_bits_in_irq: (!b->bm) in > > drivers/block/drbd/drbd_bitmap.c:1028 > > [410705.130527] drbd1: drbd_bm_set_bits_in_irq: (!b->bm) in > > drivers/block/drbd/drbd_bitmap.c:1028 > > [410705.130760] drbd1: drbd_bm_set_bits_in_irq: (!b->bm) in > > drivers/block/drbd/drbd_bitmap.c:1028 > > [410706.518114] drbd1: drbd_bm_set_bits_in_irq: (!b->bm) in > > drivers/block/drbd/drbd_bitmap.c:1028 > > [410706.522085] drbd1: drbd_bm_set_bits_in_irq: (!b->bm) in > > drivers/block/drbd/drbd_bitmap.c:1028 > > [root at printemps modules]# cat /proc/drbd > > version: 8.0.7 (api:86/proto:86) > > GIT-hash: cf14288833afe95db396075f8530a5960d29e498 build by > > root at printemps, 2007-12-05 18:13:36 > > 0: cs:Connected st:Primary/Secondary ds:Diskless/UpToDate C r--- > > ns:205554 nr:0 dw:0 dr:0 al:0 bm:0 lo:0 pe:0 ua:0 ap:0 > > 1: cs:Connected st:Primary/Secondary ds:Diskless/UpToDate C r--- > > ns:9411157 nr:0 dw:0 dr:0 al:0 bm:0 lo:0 pe:0 ua:0 ap:0 > > > > I've attached a "relevent" kern.log... > > which was even as bzip2 one Megabyte, > really too big for this list. > > anyways, first find out why you are Diskless. Don't have a clue! The complete partition table on /dev/c0d1 simply disapeared ??? and I don't know since when? I had to remove each logical volumes then the volume group to recreate a new primary partition then vg then lv's... And now I'm finally resynching the both drbd containers... Maybie the problem originally came from there? > then, grep in your log for "bitmap: failed to vmalloc" > maybe you are diskless because the allocation of the bitmap failed? > > also, you could check wether upgrading to 8.0.8 changes behaviour. I've upgraded one system to use version 8.0.8. I'll check. Although it did not happend again on any of the two systems and I'm not even sure if it ever happened anywhere else then on the diskless node? > > we probably want to make that error message less noisy, > anyways, e.g. by a rate-limit. > - vin