Docker update from 1.3.x not possible
Issue #1261
resolved
Docker Container update from 1.3.x to 1.4.2 is not possible.
With the following error the container goes into an error:
sed: can't read /etc/piler/manticore.conf: No such file or directory
sed: can't read /etc/piler/manticore.conf: No such file or directory
sed: can't read /etc/piler/manticore.conf: No such file or directory
sed: can't read /etc/piler/manticore.conf: No such file or directory
sed: can't read /etc/piler/manticore.conf: No such file or directory
sed: can't read /etc/piler/manticore.conf: No such file or directory
sed: can't read /etc/piler/manticore.conf: No such file or directory
Comments (27)
-
repo owner
-
reporter
/etc/piler/MANTICORE is not present
-
reporter
root@ubuntu-xxxx:/var/lib/docker/volumes/piler-docker_piler_etc/_data# ls -lah
total 76K
drwxr-xr-x 3 root root 4.0K Dec 18 12:34 .
drwx-----x 3 root root 4.0K Dec 18 12:34 ..
-rw-r--r-- 1 root 1000 379 Feb 23 2021 config-site.dist.php
-rw-r--r-- 1 root root 2.9K Dec 18 12:34 config-site.php
-rw-r--r-- 1 root root 762 Dec 18 12:34 config-site.php.bak
-rw------- 1 1000 1000 111 Dec 18 12:34 .my.cnf
-rw------- 1 1000 1000 1.2K Dec 18 12:34 piler.conf
-rw------- 1 root root 1.2K Dec 18 12:34 piler.conf.bak
-rw-r----- 1 root 1000 1.2K Feb 23 2021 piler.conf.dist
-rw------- 1 1000 1000 56 Dec 18 12:34 piler.key
-rw-r--r-- 1 root root 2.6K Dec 18 12:34 piler-nginx.conf
-rw-r--r-- 1 root root 2.5K Feb 23 2021 piler-nginx.conf.dist
-rw------- 1 1000 1000 5.2K Dec 18 12:34 piler.pem
drwxr-xr-x 2 1000 1000 4.0K Dec 18 12:34 sites
-rwxr-xr-x 1 root root 7.4K Dec 18 12:34 sphinx.conf
-rwxr-xr-x 1 root 1000 7.4K Feb 23 2021 sphinx.conf.dist
-
repo owner
You are right. I recommend you to create your dockerfile featuring sphinxsearch. The 1.4.2 tagged image now defaults to manticore.
-
reporter
Is it not possible to upgrade from 1.3.x here and switch to manticore on upgrade?
-
repo owner
I don’t follow these 2 updates. The current 1.4.2 image ships manticore and is wired to use it. You already have the index data managed by sphinx, so it makes sense to keep using it. To do that you need a different dockerfile that ships with sphinxsearch. If you decided to switch from sphinx to manticore, then you can use the 1.4.2 image as it is, just be sure to get manticore.conf, and reindex everything.
-
reporter
What would be the best way to backup all data from 1.3.12, completely rebuild the container with 1.4.2 and import the backed up data from 1.3.12 again.
I think it eight in the long run no sense to pursue Sphinx further.
Can you please explain how I get the data imported?
-
repo owner
In this case you may use the current 1.4.2 image. Create /etc/piler/manticore.conf file. Stop sphinxsearch. Create the initial index files using su -c ‘/usr/bin/indexer --all --config /etc/piler/manticore.conf' piler. Start manticore searchd, then reindex everything. It’s wise to create backups of the 2 piler related volumes.
-
reporter
I don't quite understand how to implement this.
Currently running 1.3.12 in Docker.
An update is not possible, because the 1.4.2 does not start afterwards.I would like to backup all data from the 1.3.12 container and then delete the container.
I also want to delete all volumes and then completely rebuild the 1.4.2.What data do I need to backup in the volumes so that no archived mails are lost and what is the best way to backup for this?
-
repo owner
Did you create manticore.conf on the etc volume?
-
reporter
After the container update from 1.3.12 to 1.4.2 I also tried to create a manticore.conf.
Here is the Log from the crash when starting the container
piler | Database piler | information_schema piler | piler piler | DEBUG: mysql is ready piler | rsyslogd: imklog: cannot open kernel log (/proc/kmsg): Operation not permitted. piler | rsyslogd: activation of module imklog failed [v8.2112.0 try https://www.rsyslog.com/e/2145 ] piler | * Starting periodic command scheduler cron piler | ...done. piler | * Starting nginx nginx piler | ...done. piler | DEBUG: main1.spp does not exist, creating index files piler | Manticore 5.0.2 348514c86@220530 dev (columnar 1.15.4 2fef34e@220522) (secondary 1.15.4 2fef34e@220522) piler | Copyright (c) 2001-2016, Andrew Aksyonoff piler | Copyright (c) 2008-2016, Sphinx Technologies Inc (http://sphinxsearch.com) piler | Copyright (c) 2017-2022, Manticore Software LTD (https://manticoresearch.com) piler | piler | using config file '/etc/piler/manticore.conf'... piler | indexing index 'main1'... piler | FATAL: failed to open /var/piler/manticore/main1.spl: No such file or directory, will not index. Try --rotate option. piler exited with code 1The
manticore.confand the file MANTICORE I created manually and inserted the content according to the source code.
Shouldn't the creation of the files be done when creating the container?Maybe you can fix something here with a version 1.4.3, so that the files are created correctly.
Even with an update from 1.3.12 this should not be a problem in a Conatiner or. The volumes exit already all and existing files could be patched with the container start automatically.
-
repo owner
No need to fix 1.4.2, you need to create /var/piler/manticore dir on the volume manually (pay attention to ownership and permissions). Then try again.
-
reporter
And the other Error with rsyslogd?
piler | rsyslogd: imklog: cannot open kernel log (/proc/kmsg): Operation not permitted. piler | rsyslogd: activation of module imklog failed [v8.2112.0 try https://www.rsyslog.com/e/2145 ]
-
reporter
I have now tested the following:
- manticore.conf created in etc
- MANTICORE created in etc
- manticore directory created in var
The container starts, but the webinterface is not reachable (404 Not Found)
What else can I do?
Enclosed the log
piler | Database piler | information_schema piler | piler piler | DEBUG: mysql is ready piler | DEBUG: metadata table exists piler | rsyslogd: imklog: cannot open kernel log (/proc/kmsg): Operation not permitted. piler | rsyslogd: activation of module imklog failed [v8.2112.0 try https://www.rsyslog.com/e/2145 ] piler | * Starting periodic command scheduler cron piler | ...done. piler | * Starting nginx nginx piler | ...done. piler | starting searchd . . . piler | [Mon Dec 19 08:24:17.667 2022] [57] using config file '/etc/piler/manticore.conf' (10645 chars)... piler | Manticore 5.0.2 348514c86@220530 dev (columnar 1.15.4 2fef34e@220522) (secondary 1.15.4 2fef34e@220522) piler | Copyright (c) 2001-2016, Andrew Aksyonoff piler | Copyright (c) 2008-2016, Sphinx Technologies Inc (http://sphinxsearch.com) piler | Copyright (c) 2017-2022, Manticore Software LTD (https://manticoresearch.com) piler | piler | starting daemon version '5.0.2 348514c86@220530 dev (columnar 1.15.4 2fef34e@220522) (secondary 1.15.4 2fef34e@220522)' ... piler | listening on 127.0.0.1:9312 for sphinx and http(s) piler | listening on 127.0.0.1:9306 for mysql piler | Manticore 5.0.2 348514c86@220530 dev (columnar 1.15.4 2fef34e@220522) (secondary 1.15.4 2fef34e@220522) piler | Copyright (c) 2001-2016, Andrew Aksyonoff piler | Copyright (c) 2008-2016, Sphinx Technologies Inc (http://sphinxsearch.com) piler | Copyright (c) 2017-2022, Manticore Software LTD (https://manticoresearch.com) piler | piler | precaching index 'main1' piler | precaching index 'main2' piler | precaching index 'main3' piler | precaching index 'main4' piler | precaching index 'dailydelta1' piler | precaching index 'delta1' piler | precaching index 'tag1' piler | precaching index 'note1' piler | Manticore 5.0.2 348514c86@220530 dev (columnar 1.15.4 2fef34e@220522) (secondary 1.15.4 2fef34e@220522) piler | Copyright (c) 2001-2016, Andrew Aksyonoff piler | Copyright (c) 2008-2016, Sphinx Technologies Inc (http://sphinxsearch.com) piler | Copyright (c) 2017-2022, Manticore Software LTD (https://manticoresearch.com) piler | piler | starting piler-smtp . . . piler | starting piler . . .
-
repo owner
Check the nginx error logs in the container.
-
reporter
I have now found a solution.
Not nice, but this way I could update the container
docker-compose down rm -rf /var/lib/docker/volumes/piler-docker_piler_var/_data/www mkdir -p /var/lib/docker/volumes/piler-docker_piler_var/_data/manticore chown 1000:1000 /var/lib/docker/volumes/piler-docker_piler_var/_data/manticore docker volume rm piler-docker_piler_etc docker-compose up --force-recreate --build -dFurthermore I created a volume for "/var/lib/www" and restored the user settings in etc after the upgrade
It would be awesome if you could make an update to the Docker container that takes these things into account during the build.
As it currently is with the 1.4.2, no one gets a container update to run
-
reporter
I would even go so far with the volumes that all data such as store etc. is packed into its own volumes.
Think that increases the data security during upgrades.
So you can quickly delete a volume and recreate it with a build. -
repo owner
You are right, I’ll fix the volume settings to /var/piler/store, like https://bitbucket.org/jsuto/piler/commits/b072f4bd949fd4efa82e467c234ab430e3eedfc7
-
reporter
Don't other directories in /var/piler need to be persistent?
What about the imap, manticore, sphinx, and stat directories?
Won't data be lost there on a reboot?root@ubuntu-22-test:/var/lib/docker/volumes/piler-docker_piler_var/_data# ls -lah total 40K drwxr-xr-x 10 root root 4.0K Dec 20 13:34 . drwx-----x 3 root root 4.0K Dec 20 10:53 .. drwx--x--x 2 1000 1000 4.0K Feb 23 2021 error drwx------ 2 1000 1000 4.0K Feb 23 2021 imap drwxr-xr-x 2 1000 1000 4.0K Dec 20 15:35 manticore drwx------ 2 1000 1000 4.0K Dec 20 10:53 sphinx drwxr-xr-x 2 1000 1000 4.0K Dec 20 11:05 stat drwx------ 3 1000 1000 4.0K Dec 20 10:53 store drwx--x--x 4 1000 1000 4.0K Dec 20 10:53 tmp drwxr-xr-x 2 root root 4.0K Dec 20 13:34 www
-
reporter
Also in "/etc/piler" the question would be if you need the complete folder persistent, or maybe you just create volumes for single config files?
I think the less you bind in persistent directories, the less problems arise with updates
-
repo owner
You are right. The manticore dir should be persistent as well. The error dir holds received, but failed to processed emails. The tmp dir contains temp files waiting to be processed.
-
reporter
My approach would be to create the following volumes to make the container flexible for future update.
for var/piler:
imap ??? manticore stat ??? storeand for /etc/piler only files on a Volume
config-site.php manticore.conf .my.cnf ??? piler.confWhat do you think about this? Did I forget something that also needs to be persistent?
Or do I have possibly too much in it? -
I think also nginx, cron and other services should be separate containers and run with compose + there should be an example docker-compose.yml - @Janos SUTO I’ll try to make an PR…
-
repo owner
The docker stuff in the source tree is meant as a starting point, a guideline how to run piler in a container. I’m not sure why you want to move the cron stuff to a dedicated container, I’d rather move manticore and the related cron scripts to a separate container. Anyway, feel free to optimize it to your needs.
-
Hm… support for bind-mount volumes also is broken - as I can see You are assuming that volume is copied what is not the case with bind-mount’ed volumes.
-
repo owner
The /var/piler/store and /var/piler/manticore dirs are empty, so you should be able to bind mount them just fine.
-
repo owner
- changed status to resolved
I consider the issue resolved. Let me know if you have further input to discuss.
- Log in to comment
Try removing /etc/piler/MANTICORE