[Sahara] GET request with big cluster can take more than a minute
Bug #1545039 reported by
Evgeny Sikachev
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Mirantis OpenStack |
Invalid
|
High
|
Evgeny Sikachev | ||
| 9.x |
Invalid
|
High
|
Evgeny Sikachev | ||
Bug Description
ENVIRONMENT: MOS 8.0 ISO 529(RC1)
If sahara has the big cluster, for example, CDH with 1 master, 1 manager and 300 workers, GET requests can take more than a minute. You can update timeout for connection to database in /etc/sahara/
This bug reproduced via UI and CLI sahara client
| Changed in mos: | |
| milestone: | 8.0 → 8.0-updates |
| importance: | Undecided → High |
Revision history for this message
| Roman Podoliaka (rpodolyaka) wrote | #1 |
| Changed in mos: | |
| status: | New → Incomplete |
Revision history for this message
| Roman Podoliaka (rpodolyaka) wrote | #2 |
| Changed in mos: | |
| assignee: | MOS Sahara (mos-sahara) → Evgeny Sikachev (esikachev) |
Revision history for this message
| Evgeny Sikachev (esikachev) wrote | #3 |
Revision history for this message
| Evgeny Sikachev (esikachev) wrote | #4 |
| Changed in mos: | |
| status: | Incomplete → Invalid |
To post a comment you must log in.
I suggest we take a look at this on Sahara side, maybe we are doing something very inefficiently.
But overall, with synchronous requests we are going to hit similar problems here and there as we scale. The fundamental solution should be using of pagination / asynchronous processing of requests / etc.
Vitaly, could you please elaborate what specific GET requests times out?
By the way, you would also need to tweak the timeout in haproxy.