search

google-code-export / lusca-cache

Automatically exported from code.google.com/p/lusca-cache
0 stars 0 forks source link

COSS doesn't work (swaplog not created?), latest HEAD #113

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
Corresponding lines in config

cache_dir coss /dev/sdb1 16000 block-size=4096 membufs=256 max-size=65536
cache_swap_state /cache1/squid/%s

from cache.log
2010/05/28 15:01:36| Rebuilding COSS storage in /dev/sdb1 (DIRTY)
2010/05/28 15:01:36| COSS[2216]: /dev/sdb1: Rebuilding 0.00% complete (0
out of 16000 stripes)
2010/05/28 15:01:50| COSS[2216]: /dev/sdb1: Rebuilding 5.00% complete (800
out of 16000 stripes)
2010/05/28 15:02:03| COSS[2216]: /dev/sdb1: Rebuilding 10.00% complete
(1600 out of 16000 stripes)
2010/05/28 15:02:17| COSS[2216]: /dev/sdb1: Rebuilding 15.00% complete
(2400 out of 16000 stripes)
2010/05/28 15:02:30| COSS[2216]: /dev/sdb1: Rebuilding 20.00% complete
(3200 out of 16000 stripes)
2010/05/28 15:02:43| COSS[2216]: /dev/sdb1: Rebuilding 25.00% complete
(4000 out of 16000 stripes)
2010/05/28 15:02:57| COSS[2216]: /dev/sdb1: Rebuilding 30.00% complete
(4800 out of 16000 stripes)
2010/05/28 15:03:10| COSS[2216]: /dev/sdb1: Rebuilding 35.00% complete
(5600 out of 16000 stripes)
2010/05/28 15:03:24| COSS[2216]: /dev/sdb1: Rebuilding 40.00% complete
(6400 out of 16000 stripes)
2010/05/28 15:03:37| COSS[2216]: /dev/sdb1: Rebuilding 45.00% complete
(7200 out of 16000 stripes)
2010/05/28 15:03:51| COSS[2216]: /dev/sdb1: Rebuilding 50.00% complete
(8000 out of 16000 stripes)
2010/05/28 15:04:05| COSS[2216]: /dev/sdb1: Rebuilding 55.00% complete
(8800 out of 16000 stripes)
2010/05/28 15:04:18| COSS[2216]: /dev/sdb1: Rebuilding 60.00% complete
(9600 out of 16000 stripes)
2010/05/28 15:04:32| COSS[2216]: /dev/sdb1: Rebuilding 65.00% complete
(10400 out of 16000 stripes)
2010/05/28 15:04:46| COSS[2216]: /dev/sdb1: Rebuilding 70.00% complete
(11200 out of 16000 stripes)
2010/05/28 15:05:00| COSS[2216]: /dev/sdb1: Rebuilding 75.00% complete
(12000 out of 16000 stripes)
2010/05/28 15:05:14| COSS[2216]: /dev/sdb1: Rebuilding 80.00% complete
(12800 out of 16000 stripes)
2010/05/28 15:05:28| COSS[2216]: /dev/sdb1: Rebuilding 85.00% complete
(13600 out of 16000 stripes)
2010/05/28 15:05:41| COSS[2216]: /dev/sdb1: Rebuilding 90.00% complete
(14400 out of 16000 stripes)
2010/05/28 15:05:55| COSS[2216]: /dev/sdb1: Rebuilding 95.00% complete
(15200 out of 16000 stripes)
2010/05/28 15:06:03| COSS: /dev/sdb1: Rebuild Completed
2010/05/28 15:06:04| storeCossWriteMemBufDone: got failure (-6)
2010/05/28 15:06:05| storeCossWriteMemBufDone: got failure (-6)
2010/05/28 15:06:06| storeCossWriteMemBufDone: got failure (-6)
and also
2010/05/28 15:12:04| WARNING: failed to unpack meta data
2010/05/28 15:12:05| WARNING: failed to unpack meta data
2010/05/28 15:12:10| WARNING: failed to unpack meta data
2010/05/28 15:12:11| WARNING: failed to unpack meta data
2010/05/28 15:12:12| WARNING: failed to unpack meta data
2010/05/28 15:12:12| WARNING: failed to unpack meta data
2010/05/28 15:12:13| WARNING: failed to unpack meta data
2010/05/28 15:12:13| WARNING: failed to unpack meta data
2010/05/28 15:12:14| WARNING: failed to unpack meta data
2010/05/28 15:12:14| WARNING: failed to unpack meta data
2010/05/28 15:12:14| WARNING: failed to unpack meta data

SUPERPROXY /cache1/squid # sfdisk -l /dev/sdb

Disk /dev/sdb: 8924 cylinders, 255 heads, 63 sectors/track
Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0

   Device Boot Start     End   #cyls    #blocks   Id  System
/dev/sdb1          0+   1992    1993-  16008741   83  Linux
/dev/sdb2       1993    8923    6931   55673257+  83  Linux
/dev/sdb3          0       -       0          0    0  Empty
/dev/sdb4          0       -       0          0    0  Empty

SUPERPROXY /cache1/squid # ls -la /dev/sdb*
brw-rw----    1 root     root       8,  16 May 28  2010 /dev/sdb
brw-rw----    1 squid    root       8,  17 May 28 15:13 /dev/sdb1
brw-rw----    1 root     root       8,  18 May 28  2010 /dev/sdb2

Original issue reported on code.google.com by nuclear...@gmail.com on 28 May 2010 at 12:13

GoogleCodeExporter commented 9 years ago 
did some debug

2010/05/28 15:23:16| aioWrite: FD 12: disk start: 16396582912, size 1048576
2010/05/28 15:23:16| storeCossWriteMemBufDone: got failure (-6)
2010/05/28 15:23:17| aioWrite: FD 12: disk start: 16397631488, size 1048576
2010/05/28 15:23:17| storeCossWriteMemBufDone: got failure (-6)

Original comment by nuclear...@gmail.com on 28 May 2010 at 12:24

GoogleCodeExporter commented 9 years ago 
Fixed by setting size to 15000
Weird... maybe some bug hidden there, but seems it is not critical issue.

Original comment by nuclear...@gmail.com on 28 May 2010 at 12:48

GoogleCodeExporter commented 9 years ago 
Howdy!

So:

Writing to /dev/sdb - it's likely the COSS header confuses things and thus 
/dev/sdb1
looks like a screwed up partition.

I'm not sure why 16000 didn't work but 15000 did. Weird. I dont' have a 1k 
block size
disk to test, sorry.

Original comment by adrian.c...@gmail.com on 28 May 2010 at 2:22

GoogleCodeExporter commented 9 years ago 
I am not using whole HDD, just small partition (first one) (because ext4 on big
amount of small files - sucks, and coss works much better).

it's just my mess with IEC and non-IEC values. I check now:

   Device Boot Start   End    MiB    #blocks   Id  System
/dev/sdb1         0+ 15633- 15634-  16008741   83  Linux

Original comment by nuclear...@gmail.com on 28 May 2010 at 2:34