sqlite_step error looping causing 100%CPU usage

[rasos]

Symptom: 154% CPU, growing to 30% memory usage, logfile grows to 5 GB within half an hour
Environment: ownloud client 1.2.5 on Ubuntu 12.04 LTS 64bit
Syncing worked fine before the update to 1.2.5.

It seems that sqlite is not responding properly during the local walk.

05-07 10:49:51:165 _csync_detect_update: file: zzzzzzz.jpg - hash 16769791247635479965, mtime: 1324303105
05-07 10:49:51:165 csync_statedb_query: sqlite_step error: disk I/O error - on query: SELECT * FROM metadata WHERE phash='-1676952826074071651'
05-07 10:49:51:165 csync_statedb_get_stat_by_hash: No result record found for phash = 16769791247635479965
05-07 10:49:51:165 csync_statedb_query: sqlite_step error: disk I/O error - on query: SELECT * FROM metadata WHERE inode='1795706'

here is the error message repeating endlessly:
cat owncloudlog07May2013.log | grep sqlite_step | more

05-07 10:29:22:478 csync_statedb_query: sqlite_step error: disk I/O error - on query: SELECT md5 FROM metadata WHERE phash='-3687424326552511022' AND modtime=1324303086
05-07 10:29:22:478 csync_statedb_query: sqlite_step error: disk I/O error - on query: SELECT * FROM metadata WHERE phash='-3687424326552511022'
05-07 10:29:22:478 csync_statedb_query: sqlite_step error: disk I/O error - on query: SELECT * FROM metadata WHERE inode='1794531'
05-07 10:29:22:478 csync_statedb_query: sqlite_step error: disk I/O error - on query: SELECT md5 FROM metadata WHERE phash='-2530360419868093365' AND modtime=1324303086
05-07 10:29:22:478 csync_statedb_query: sqlite_step error: disk I/O error - on query: SELECT * FROM metadata WHERE phash='-2530360419868093365'
05-07 10:29:22:478 csync_statedb_query: sqlite_step error: disk I/O error - on query: SELECT * FROM metadata WHERE inode='1795188'

There are some reports on the owncloud forum about 100% CPU usage as well, so this may be related:
https://forum.owncloud.org/viewtopic.php?f=14&t=10761
http://forum.owncloud.org/viewtopic.php?f=14&t=10472

Should I start from scratch with an empty sqlite DB? How?

[ghost]

Hi,

this is already reported here:

#591

[rasos]

Update: the high mem and CPU load disappeared, when all sync folders were available again on the remote machine. One of the shared root folders with 100k+ files was not accessible due to a permission issue, which is fixed now.

[dragotin]

Hmm, so I think i can close this.