Page MenuHomeDevCentral

Evaluate if Zammad wouldn't be useful
Open, NormalPublic


One of the OTRS founder launched an open source Ruby server application described as a Zendesk alternative.

It allows to interact by mail, phone, chat.

The UI looks good:

Event Timeline


NOTE: An important thing to stress upon: there is a migration path from OTRS to Zammad.

Happy about

It's promising. UI is beautiful. Server mail configuration is awesome, polling is reactive (I had the postfix console opened to check what it does, that includes automatic testing through

It manages multiple organizations and has a good REST API.

Less happy about

But technically, I noticed some glitches:

  • branding: images aren't served by default by the Docker compose installation
  • elasticsearch: some queries try to post to http://elasticsearch:9200/...
  • typos: 'text modues'
  • views: where is "all ticket views"?

Not happy about

When we test, it's still somewhat immature:

  • outgoing mail worked out of the box, but for incoming mail, it prints the number of mail in the inbox when we configure it, but doesn't process them
  • web forms require jQuery, designer prints obvious jQuery code with $ selector, but doesn't document it's explicitly required (it requires jQuery 2, not 3 by the way)
  • web forms answer 'Sorry, you look like an robot!' for small messages
  • web forms answer 'Thank you for your inquiry (#67004)! We'll contact you as soon as possible.' for processed messages
  • UI asked me to enable notifications, so why I don't get a notification for this #67004 ticket (nor any JS error about that)?
  • Some 404 errors on /api paths


I reported these two issues upstream:

Memory required

4 Gb are advertised (2 Gb for the server, 2 Gb for Elasticsearch) but when tried on my desktop, I got this 1350 Mb memory usage:

$ docker stats ...
CONTAINER                                  CPU %               MEM USAGE / LIMIT       MEM %               NET I/O               BLOCK I/O             PIDS
zammaddockercompose_nginx_1                0.00%               4.5 MiB / 5.574 GiB     0.08%               4.889 MB / 15.97 MB   10.83 MB / 0 B        2
zammaddockercompose_zammad-railsserver_1   0.01%               266.8 MiB / 5.574 GiB   4.67%               23.23 MB / 16.54 MB   9.949 MB / 28.67 kB   12
zammaddockercompose_zammad-scheduler_1     0.06%               218.1 MiB / 5.574 GiB   3.82%               29.32 MB / 7.845 MB   13.59 MB / 0 B        10
zammaddockercompose_zammad-websocket_1     0.05%               103.8 MiB / 5.574 GiB   1.82%               729.8 kB / 142.1 kB   13.59 MB / 0 B        3
zammaddockercompose_elasticsearch_1        0.13%               678.1 MiB / 5.574 GiB   11.88%              395 kB / 29.6 kB      8.249 MB / 208.9 kB   48
zammaddockercompose_zammad_1               --                  -- / --                 --                  -- / --               -- / --               --
zammaddockercompose_postgresql_1           0.00%               79.13 MiB / 5.574 GiB   1.39%               20.59 MB / 49.86 MB   15.02 MB / 232.9 MB   20

For Elasticsearch, it started at 2 Gb and shrink to this 678.1 MiB once stabilized.


Deploy OTRS 5.

Watch and see if Zammad gain some traction in 6-18 months. If so, migrate from OTRS to Zammad.

By the way, Dwellers doesn't have enough RAM. We also clearly need to get more RAM capacity for the Nasqueron Docker engine, and then it will be okay to migrate.

we're at the +18 months moment and Zammad is still in active development, with a new release this December.

The issues noted above looks like early development glitches, and we can reasonably expect them to be few now. Time to test again.

A new version 3.2.0 has been released 2019-12-03 and contains an interesting part:

Technical notes (for self hosted):

Performance improvements:
CPU usage - Several improvements have been made to significantly reduce the CPU usage of the scheduler (further improvements will be made in future versions).
Memory footprint - A new environment variable for Ruby memory allocation has significantly reduced memory consumption (in optimal cases, memory consumption is reduced by 38% in long-term operation).

Twitter support discontinuation

Similarly to T1795, we're not supporting Twitter features any longer if we adopt Zammad.

Zammad 6 has been released in June.

It offers a mobile view and detect duplicates.