You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by "Kahn, Peter" <pk...@connected.com> on 2005/09/26 23:06:41 UTC
Hotcopy failure
I have a repository on fedora core 4 with svn 1.2.1 and Berk 4.3. This
db was created in the svn 0.6-0.7 time frame and hasn't been through a
dump/load cycle since. I am using http and svnserve to access my
repository. My repository I about 6.4 GB in size.
For the past two days when I try to hotcopy my repository (even when I
do not have any service accessing it) I get this error:
svnadmin: DB_RUNRECOVERY: Fatal error, run database recovery
svnadmin: bdb: Logging region out of memory; you may need to
increase its size
svnadmin: bdb: Recovery function for LSN 18565 238216 failed
svnadmin: bdb: PANIC: Cannot allocate memory
svnadmin: bdb: PANIC: fatal region error detected; run recovery
I am planning to do a dump/reload soon to allow me to go up to svn123,
because I cannot without doing so for some reason. This being said, I'd
like to be able to do this at my leisure. The fact that I cannot backup
has my quite worried about the state of my db.
When I run recover it finishes quickly and says there are no errors....
Could it be that my DB_CONFIG should be altered? I have 2GB of ram on
the box, so I have no problem upping any values.
Please let me know if anyone has any clues or suggestions. Thanks...
My DB_CONFIG file
# This is the configuration file for the Berkeley DB environment
# used by your Subversion repository.
# You must run 'svnadmin recover' whenever you modify this file,
# for your changes to take effect.
#set_cachesize 1048576
### Lock subsystem
#
# Make sure you read the documentation at:
#
# http://www.sleepycat.com/docs/ref/lock/max.html
#
# before tweaking these values.
set_lk_max_locks 2000
set_lk_max_lockers 2000
set_lk_max_objects 2000
### Log file subsystem
#
# Make sure you read the documentation at:
#
# http://www.sleepycat.com/docs/api_c/env_set_lg_bsize.html
# http://www.sleepycat.com/docs/api_c/env_set_lg_max.html
# http://www.sleepycat.com/docs/ref/log/limits.html
#
# Increase the size of the in-memory log buffer from the default
# of 32 Kbytes to 256 Kbytes. Decrease the log file size from
# 10 Mbytes to 1 Mbyte. This will help reduce the amount of disk
# space required for hot backups. The size of the log file must be
# at least four times the size of the in-memory log buffer.
#
# Note: Decreasing the in-memory buffer size below 256 Kbytes
# will hurt commit performance. For details, see this post from
# Daniel Berlin <da...@dberlin.org>:
#
# http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgId=161960
set_lg_bsize 262144
set_lg_max 1048576
#
# If you see "log region out of memory" errors, bump lg_regionmax.
# See http://www.sleepycat.com/docs/ref/log/config.html and
# http://svn.haxx.se/users/archive-2004-10/1001.shtml for more.
set_lg_regionmax 131072
#
# The default cache size in BDB is only 256k. As explained in
# http://svn.haxx.se/dev/archive-2004-12/0369.shtml, this is too
# small for most applications. Bump this number if "db_stat -m"
# shows too many cache misses.
set_cachesize 0 1048576 1
#
# Disable fsync of log files on transaction commit. Read the
# documentation about DB_TXN_NOSYNC at:
#
# http://www.sleepycat.com/docs/api_c/env_set_flags.html
#
# [requires Berkeley DB 4.0]
# set_flags DB_TXN_NOSYNC
#
# Enable automatic removal of unused transaction log files.
# Read the documentation about DB_LOG_AUTOREMOVE at:
#
# http://www.sleepycat.com/docs/api_c/env_set_flags.html
#
# [requires Berkeley DB 4.2]
#set_flags DB_LOG_AUTOREMOVE
--
Peter Kahn
pkahn@connected.com
TipicIM: pkahn@imserver
Iron Mountain Digital
PackReq Described:
http://svn/svn/repos/Welcome_To_SVN/buildFarmTutorial.html#Packager
Network Map:
http://teamconnected/Engineering/Eng2_Intranet/release_eng/map.htm
Build Machine Info: http://ictinus/openwiki/ow.asp?BuildMachineRoadWork
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: Hotcopy failure
Posted by kf...@collab.net.
"Kahn, Peter" <pk...@connected.com> writes:
> I have a repository on fedora core 4 with svn 1.2.1 and Berk 4.3. This
> db was created in the svn 0.6-0.7 time frame and hasn't been through a
> dump/load cycle since. I am using http and svnserve to access my
> repository. My repository I about 6.4 GB in size.
>
> For the past two days when I try to hotcopy my repository (even when I
> do not have any service accessing it) I get this error:
> svnadmin: DB_RUNRECOVERY: Fatal error, run database recovery
> svnadmin: bdb: Logging region out of memory; you may need to
> increase its size
> svnadmin: bdb: Recovery function for LSN 18565 238216 failed
> svnadmin: bdb: PANIC: Cannot allocate memory
> svnadmin: bdb: PANIC: fatal region error detected; run recovery
>
>
> I am planning to do a dump/reload soon to allow me to go up to svn123,
> because I cannot without doing so for some reason. This being said, I'd
> like to be able to do this at my leisure. The fact that I cannot backup
> has my quite worried about the state of my db.
>
> When I run recover it finishes quickly and says there are no errors....
> Could it be that my DB_CONFIG should be altered? I have 2GB of ram on
> the box, so I have no problem upping any values.
>
> Please let me know if anyone has any clues or suggestions. Thanks...
http://subversion.tigris.org/svn_1.2_releasenotes.html#bdb-upgrade
might have some useful information for you.
-Karl
--
www.collab.net <> CollabNet | Distributed Development On Demand
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org