Consumer: Cannot connect to redis://:**@localhost:6379//: Client sent AUTH, but no password is set

I installed Koa from the master branch on 17th March and the installation completed successfully with this message
PLAY RECAP *********************************************************************
localhost : ok=710 changed=476 unreachable=0 failed=0 skipped=1465 rescued=0 ignored=5

Installation finished at 2021-03-17 20:12:08

when I issued /edx/bin/supervisorctl command, there’s no notifier-scheduler or notifier-celery-workers service. When I read release note for Koa, it seems the notifier has been removed as well.

It seems that rabbitmq also removed, as there is no rabbitmq config file under
/etc/rabbitmq/rabbitmq-env.conf

But in the edx.log I still have this error message:
Mar 17 19:41:27 localhost [service_variant=lms][celery.worker.consumer.consumer][env:sandbox] ERROR [localhost 59938] [user None] [ip None] [consumer.py:428] – consumer: Cannot connect to redis://:**@localhost:6379//: Client sent AUTH, but no password is set.

I have checked that /etc/redis/redis.conf , requirepass directive is commented out and the password is the same with the password in my-passwords.yml.

I read this thread here:

But I think should not be the same cause, because I just use the recent master branch, where the patch should already be committed.

Please advise. Thank you in advance

@chr1sk did you ever figure this one out? I’m hitting the same error. Especially odd since it’s a copy of another machine with what I believe to be the same configuration (modified some storage values, but nothing related to celery/redis).
Additionally, I’m seeing edxapp_worker:lms_default_1 in ‘STOPPING’ state. All other workers are fine.

I ended up finding a solution here that worked for me.
What led me to it was I found that /var/lib/redis/dump.rdb was about 200x larger on my server with the error than a few others I looked at. (still only about 1.1 MB, but these files are apparently usually not very big).
In addition, if I went into redis command line (redis-cli) and monitored it (monitor), I wasn’t seeing any entries for “edx.lms.core.default”. They all say “edx.lms.core.high” or “edx.lms.core.high_mem”.

So I flushed all redis from redis command line with FLUSHALL
(From what I’ve read and observed, this is safe. But I took a backup of my dump.rdb file and the server overall just in case).
And then I restarted all, but that didn’t fix it. So I rebooted the whole server. And that fixed it.

Now redis is running normally. my edxapp_worker:lms_default_1 has a status of RUNNING and by dump.rb file is much more reasonable:
image

I’m not 100% sure what caused this. But this machine is very much under development so it may very well have been something we did to it at some point that tripped it up. Will observe and hope it stays stable at this point.

1 Like

@chr1sk I had the same issue which said Auth is sent but no password was set.

Issue with my setup was that there was no password set for redis and I was sending one anyway.

Try to check in /edx/etc/lms.yml if you are setting a password there or not ?

yup worked fine