Re: Cluster::restart dumping logs when stop fails

> On 7 Apr 2024, at 18:51, Daniel Gustafsson <daniel(at)yesql(dot)se> wrote:
>> On 7 Apr 2024, at 18:28, Andres Freund <andres(at)anarazel(dot)de> wrote:

>> I'm ok with printing path + some content or just the path.
>
> I think printing the last 512 bytes or so would be a good approach, I'll take
> care of it later tonight. That would be a backpatchable change IMHO.

In the end I opted for just printing the path to avoid introducing logic (at
this point in the cycle) for ensuring that a negative offset doesn't exceed the
filesize. Since it changes behavior I haven't backpatched it, but can do that
if we think it's more of a fix than a change (I think it can be argued either
way, I have no strong feelings).

--
Daniel Gustafsson