System Test, RDS Core: active PDB failure on drain work-around #394
Conversation
elenagerman
force-pushed
the
pdb-wa
branch
2 times, most recently
from
e20aad8 to
09b9dc3
Compare
|
@yprokule PTAL, this is eIP failover active PDB on Drain failure workaround |
elenagerman
force-pushed
the
pdb-wa
branch
2 times, most recently
from
87550bf to
1a0b910
Compare
elenagerman
force-pushed
the
pdb-wa
branch
3 times, most recently
from
125273a to
28f6d1f
Compare
|
|
||
| return fmt.Errorf("failed to uncordon %q due to %w", nodeName, err) | ||
| } | ||
|
|
||
| time.Sleep(15 * time.Second) | ||
|
|
||
| err = restoreActivePDBValues(pdbList) |
There was a problem hiding this comment.
this part won't be executed in case there's any preceding error (draining/uncording/etc) so more appropriate place for this will be in BeforeAll to capture initial state of PDBs and Before/AfterEach to restore them, wdyt?
There was a problem hiding this comment.
I think it's a great idea, but I doubt how I can implement it. The problem is pdbList, all info for this List was collected during set... state
There was a problem hiding this comment.
and my apologies, I slightly modify both methods:
- I took it to the upper level
- now it's suitable for all pdbs, not only has == 1 values
- now the SetMinAvailableToZeroForActivePDB method collects not only pdb objects, but original values as well
- RestoreActivePDBValues, based on this map restores original values and not just set minAvailable values = 1
There was a problem hiding this comment.
actually I can separate 'collect pdb info' into stand alone function, wait a min, I'll check...
There was a problem hiding this comment.
@yprokule no, it will not help. I still do not see how I can pass to the AfterEach state PDB list. Do you have any idea?
There was a problem hiding this comment.
@yprokule do you have any idea how to implement it?
There was a problem hiding this comment.
Sent U some draft directly. Also noticed that some PDBs have their minAvailable field reverted after it's updated by the test, this must be investigated further.
Also noticed that after node reboot it takes some time for respective pods to start so proposed an extra check here - #413
elenagerman
force-pushed
the
pdb-wa
branch
5 times, most recently
from
6fb8f73 to
22e1e75
Compare
No description provided.