You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@pulsar.apache.org by GitBox <gi...@apache.org> on 2020/02/19 23:59:41 UTC
[GitHub] [pulsar] sijie commented on issue #6283: [offloaders] Define
treshold and deletionMaxLag per ns or use the default one.
sijie commented on issue #6283: [offloaders] Define treshold and deletionMaxLag per ns or use the default one.
URL: https://github.com/apache/pulsar/issues/6283#issuecomment-588537583
Adding the notes between @KannarFr and me
```
kannarfr 3:38 PM
so do you have an opinion about the last comments i posted to https://github.com/apache/pulsar/issues/6283?
KannarFrKannarFr
#6283 [offloaders] Define treshold and deletionMaxLag per ns or use the default one.
Is your feature request related to a problem? Please describe.
The OffloadPolices per ns is just merged. But it's missing a point on values to trigger auto offloading: we want to define tresholdMaxBytes and deletionMaxLag per ns or use the default one configured.
Describe the solution you'd like
Add tresholdMaxBytes and deletionMaxLag in OffloadPolices and use them to defines triggers.
Labels
type/feature
Comments
10
<https://github.com/apache/pulsar|apache/pulsar>apache/pulsar | Feb 10th | Added by GitHub
sijieg 3:43 PM
You want this to be configured per namespace ?
3:43
correct?
3:44
oh do you mean making these two settings available in OffloadPolicies?
kannarfr 3:50 PM
yes
3:50
and when tieredstorage deletion will be done, I need to add two more configuration for this deletion in offloadpolicies (edited)
3:52
so finally offloadpolicies will contains deletionlag/threshold for bk -> tieredstorage, and another couple deletion lag/threshold for tieredstorage -> deleted
3:53
but there is architecture problem/definition to discuss and find solution how to change managedledgerimpl configuration on the fly as offloadpolicies mutate
3:54
WDYT?
3:54
because currently it's only on broker configuration so on broker boot, because of the managedledgerfactory creation
3:55
so something like one managedledgerfactory per ns (because of the deletionlag/threshold per ns) etc.
sijieg 3:56 PM
I think the first thing would be introducing a per manage ledger setting to override the factory setting?
3:56
then after that, you can map offloadpolicies to the per managed ledger setting.
3:56
does that sound like a good approach?
3:56
so two seperate changes and don’t couple them together
```
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services