The website encountered an unexpected error. Please try again later.

Could this be related to my upgrade issue #1825154: Octopus Update Killed - Help!?

I'm seeing this when running the upgrade in debug mode:

WD php: RedisException: Redis server went away in Redis->auth() (line 13 of [error]

Also, does the patch get overwritten by updates/upgrades from BOA?

Thanks!
Kevin

I manually applied the patch and redis started up and my Octopus admin access was restored. I then was able to successfully update/upgrade Baracuda and Octopus and it appears that the redis patch changes persisted.

I have not ran the tuning script on this server to know if it will overwrite the patch.

Thanks TimLeytens!

Kevin

It was a result of known issue with new Redis cache server version, which introduced some changes not backward compatible and without updating its own configuration, even when it is installed via standard package and not from sources - which is default for Squeeze, so it is probably Dotdeb folks fault.

Fix committed in HEAD: http://drupalcode.org/project/barracuda.git/commit/c935281 (plus a few following commits).

This means that BOA-2.0.3 is currently broken on Squeeze, so you should use pretty stable HEAD for upgrades or new installs.

Yes, BOA-2.0.4 will be the same as HEAD, just with the release label. We are running some final tests to fix really serious bug in a caching logic.

This is not fixed yet, since newer Redis version seems to fail randomly with error: RedisException: Redis server went away in Redis->auth()

This patch seems to fix the problem: http://drupalcode.org/project/barracuda.git/commit/2807c5d

Yet another fix required: http://drupalcode.org/project/barracuda.git/commit/dde222f

@amsri As things stand now, BOA HEAD is already much more stable than (already broken) BOA-2.0.3, and we are using it in production - yes, we are that crazy ;) but the point is that it works pretty well. But yes, we also need BOA-2.0.4 release badly, so it should be ready ASAP. However, we have yet another major issue to fix first: #1825992: Redis cache is never cleared on cron nor with drush on command line

Thanks for your work omega8cc.

I just encountered this issue also. I tried a fresh, vanilla install on Debian 6. I did know there was a critical issue that needed to be resolved before 2.0.4 was released but I wasn't aware that this also affected 2.0.3.

I have an existing VPS with BOA 2.0.3 that has been experiencing problems. I'm getting a lot of 504 errors and in octopus cron doesn't seem to be running tasks. Could that be related?

To get BOA-2.0.3 based server up again, don't use barracuda up-stable to update it, but install previous Redis version manually:

cd /var/opt
rm -f -r redis*
wget -q -U iCab http://files.aegir.cc/dev/redis-2.4.17.tar.gz
tar -xzf redis-2.4.17.tar.gz
cd redis-2.4.17
make --quiet
make --quiet install
service redis-server stop
killall -9 redis-server
rm -f /var/run/redis.pid
cd /usr/local/bin
cp -p redis-server /usr/bin/
cp -p redis-benchmark /usr/bin/
cp -p redis-cli /usr/bin/
cp -p redis-check-dump /usr/bin/
cp -p redis-check-aof /usr/bin/
mkdir -p /var/log/redis
chown redis:redis /var/log/redis
mkdir -p /var/lib/redis
chown redis:redis /var/lib/redis
service redis-server stop
killall -9 redis-server
rm -f /var/run/redis.pid
rm -f /var/lib/redis/*
cd

Now wait 10 seconds for auto-healing to start Redis automatically. Done.

But the sites cron not running issue is separate and fixed in HEAD only.

Great, thank you!

The error showed up when I ran aptitude full-upgrade, which updated redis server and associated libraries(?).
I had to...

aptitude remove redis-server
aptitude install redis-server

...and then I edited /etc/redis/redis.conf with #3 and #10 to use the new config format for redis.

Wow, running aptitude full-upgrade is a bad thing. We have tried to support this gracefully, but it always tends to hit us with unpleasant surprises, so I recommend to avoid that, and always use barracuda up-{stable|head}