First month on Taiga: 12 gaps we hit and the workarounds we built

Hi!

We work at a company that moved its backlog from ClickUp to a self-hosted Taiga 6 three weeks ago (about a month counting the preparation).

We are in our first sprint on Taiga.

Some numbers so you know the size of the use: 172 user stories, 654 tasks, 29 epics,
6 sprints, 9 members.

Since go-live the team has worked on 134 stories and 226 tasks, closed 95 tasks and recorded about 1,500 changes. The current sprint has 17 stories and 129 tasks. We also did a cleanup after the import: the epic list went from 24 to 29 (4 removed, 9 created), about 100 stories changed epic, and the tag list went from 24 to 12.

We like Taiga and plan to stay.

Below are the 12 gaps we hit, with the workaround we use for each. I put everything in one post to keep the context together, but if you prefer one topic per suggestion, just say so and I will open them separately.

# Suggestion Problem we hit Use case Our workaround
1 Configurable columns in the backlog list The backlog shows a fixed set of fields; custom fields are not visible During planning we need to see Priority and Requester next to each story without opening it A Google Sheet fed by the CSV report
2 Priority as a built-in field on user stories Only issues have priority; on stories we had to create a custom dropdown Our PO triages ~90 open stories by Urgent / High / Normal / Low Custom field “Priority”
3 Sort the backlog by field Ordering is drag-and-drop only Sort open stories by priority, epic or assignee before a planning meeting Same Google Sheet
4 Epic hierarchy (sub-epics, or one level between epic and story) Taiga has Epic > Story > Task; we came from a tool with one more level and had to flatten it “Lead intake” (work stream) contains “AI chat”, “Chat platform” and “Lead management” (features), each with its own stories Sibling epics grouped by convention (epic owner + area tag)
5 Bulk edit of user stories No way to change epic, tags or status of many stories at once in the UI Moving ~100 stories to new epics while reorganizing the epic list Scripts against the API
6 Pause e-mail notifications during bulk changes Every change notifies the people involved; there is no admin “quiet mode” One reorganization queued 117 notifications We set the notify policies to “none” directly in the database and restored them afterwards
7 Custom fields usable in filters and sorting Custom fields only appear inside the story Filter the backlog by Requester (the business person who asked for it) Google Sheet
8 Numeric “hours” field with totals per sprint and per person Estimation is points only; a custom text field is not validated or summed and does not feed the burndown Check each analyst’s allocated hours against their capacity for the sprint Custom text field plus a Google Apps Script report
9 Bug as a story type, or issues with tasks and epics Issues have no tasks or epic and stay outside the sprint flow A production bug needs tasks (reproduce, fix, test, deploy) and belongs to an epic Bugs are user stories tagged “bug”
10 Make API pagination explicit List endpoints return 30 items unless a header is sent; our client silently worked with partial lists Listing all stories of a project to check for duplicates Fixed in our client (x-disable-pagination)
11 Dark theme in the Docker deployment We could not find a way to select a dark theme in our Docker install; it seems to be available only in other setups Several people on the team work in dark mode all day None so far. If there is a setting we missed, a pointer would be great
12 Gateway should survive container recreation taiga-gateway resolves taiga-front and taiga-back once at start; after recreating those containers it keeps the old address and the site returns 502 Changing a value in .env and running “docker compose up -d” Restart taiga-gateway after every recreate; we documented it internally

Items 1, 2, 3 and 7 are really one theme: priority and custom fields as first-class citizens in the backlog. If we had to pick, that theme and item 4 would help us the most.

Happy to give more detail or test anything. Thanks for Taiga!

1 Like

On #12, that’s standard nginx behavior rather than something Taiga-specific: when proxy_pass uses a literal hostname, nginx resolves it once at startup and keeps that IP, so a recreated taiga-back or taiga-front with a new address gives 502s until the gateway restarts.

You can make the gateway re-resolve by pointing nginx at Docker’s embedded DNS and putting the upstream in a variable, for example in taiga-gateway/taiga.conf:

server {
    listen 80 default_server;
    resolver 127.0.0.11 valid=10s ipv6=off;

    location / {
        set $taiga_front http://taiga-front;
        proxy_pass $taiga_front;
        # ...existing proxy_set_header lines
    }

    location /api/ {
        set $taiga_back http://taiga-back:8000;
        proxy_pass $taiga_back;
        # ...existing proxy_set_header lines
    }
}

One caveat: with a variable and no URI part, nginx passes the original request URI unchanged. That’s fine for /, /api/ and /admin/ because the prefixes already match, but /media/ currently strips its prefix (proxy_pass http://taiga-protected:8003/), so leave that block as is or add a rewrite before switching it.

Great write-up, by the way. We think about a lot of the same questions (priority and custom fields as first-class in views) while building Novastart, an infinite canvas workspace for teams, so items 1, 3 and 7 really resonate.

Good explanation. One addition: leaving the /media/ block with a literal upstream only avoids the URI-prefix issue, not the stale-IP issue if that service is recreated. If you need /media/ to re-resolve too, use the resolver plus a variable, and add an explicit rewrite that strips the /media/ prefix before proxying so the backend still receives the same path. Otherwise your caveat is correct.