Change event_payload column size from TEXT to MEDIUMTEXT

[eduardoj]

This change was missing when the payload colum size of table events was changed.

Related to #15649.

[darix]

while I think this change is a good step in the right direction ... might i suggest a follow up PR which basically adds some exception handling so that it can skip over such events and try requests further down the list?

and maybe some influxdb metric that tells us about failed notifications.

[eduardoj]

This was a first and fast approach to solve the issue of not creating notifications.

I put this PR in draft, and will think in a better solution.

[hennevogel]

might i suggest ... some exception handling so that it can skip over such events and try requests further down the list?

Yeah delayed_job is not too smart about low level failures. I'm sure this will come up in the RCA :)

and maybe some influxdb metric that tells us about failed notifications.

We already measure everything about this, including alerts. They just fired when no one was there anymore...

https://obs-measure.opensuse.org/d/SKLo7evMk/emails?orgId=1
https://obs-measure.opensuse.org/d/N1ubohEMz/notifications?orgId=1
https://obs-measure.opensuse.org/d/zUE8_eM7k/activejob?orgId=1&refresh=5m

[darix]

This was a first and fast approach to solve the issue of not creating notifications.

I put this PR in draft, and will think in a better solution.

TBH making all the tables use the same size for the columns is good. i was just thinking how we can handle this better if similar errors show up in the future.