How to manually trigger retention policy?

[mloskot]

I have configured a simple All Events retention policy. It works, of course.
Meanwhile, I added lots of new deployments - increased amount of ingested logs.
Today, I noticed Seq is running out of disk space, so I thought I will clean it up this way:

1. Decrease retention policy period to 3 days.
2. Wait until the retention policy is triggered.

It worked, of course, but I had to wait quite some time for Seq to trigger the updated retention policy.

Is there any way, CLI command or API request, to manually trigger the policy and avoid the waiting?

Or is this https://datalust.co/docs/deleting-events the only manual way to perform clean ups in such disk space emergencies?

[nblumhardt]

Hi @mloskot,

Sending an HTTP DELETE to /api/events/signal?toDateUtc=<some ISO datetime> will perform essentially the same function as an "All events" policy; I think it's best if the datetime is aligned to a five minute boundary (I need to check if that constraint still influences the deletion strategy chosen..).

The C# API client has connection.Events.DeleteAsync() to call this, if it helps.

(Yes, any ISO 8601 timestamp is fine for that toDateUtc parameter - the name is a relic from an earlier, less-enlightened era ;-))

Hope this helps,
Nick

[mloskot]

Hi @nblumhardt,

The HTTP suggestion is what I was looking for, I'm gonna try that soon.

For the sake of convenience and interface parity, there could be something like seqcli events delete --to-date-ufc command. WDYT?

[nblumhardt]

Sounds good to me @mloskot 👍

If you're interested in having a shot at it - the seqcli search command supports a very similar set of parameters to what events delete would allow, so that would be the best to model it on (there's already a standardized DateRangeFeature used there for date range arguments, for example).

[mloskot]

Hi @nblumhardt,
I've submitted PR with seqcli delete proposal:

• feat: Add seqcli events delete command seqcli#445

I'll appreciate your review and feedback.