Merge ~bryce/ubuntu/+source/apache2:sru-lp1832182-graceful-groovy into ubuntu/+source/apache2:ubuntu/groovy-devel
Status: | Merged |
---|---|
Approved by: | Bryce Harrington on 2020-11-19 |
Approved revision: | 5b03b0873f8a139acdcdcf0704ea6558eba28785 |
Merge reported by: | Christian Ehrhardt |
Merged at revision: | 5b03b0873f8a139acdcdcf0704ea6558eba28785 |
Proposed branch: | ~bryce/ubuntu/+source/apache2:sru-lp1832182-graceful-groovy |
Merge into: | ubuntu/+source/apache2:ubuntu/groovy-devel |
Diff against target: |
98 lines (+40/-18) 2 files modified
debian/apache2ctl (+30/-18) debian/changelog (+10/-0) |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Christian Ehrhardt | 2020-11-13 | Approve on 2020-11-17 | |
Ubuntu Server Dev import team | 2020-11-13 | Pending | |
Canonical Server Team | 2020-11-13 | Pending | |
Review via email:
|
Description of the change
Backport to groovy of fix from
https:/
The start command was updated to work with systemd earlier; this extends the same fix for the 'graceful' command.
PPA: https:/
SRU: https:/
Forwarded to Debian: https:/
Christian Ehrhardt (paelzer) wrote : | # |
Christian Ehrhardt (paelzer) wrote : | # |
I remember reviewing the same fix for another release.
This one still LGTM and makes sense to avoid the direct restart breaking status detection.
Also changelog and version are ok.
The only thing I didn't get to this time is testing, have you run your PPA through upgrade tests already?
Christian Ehrhardt (paelzer) wrote : | # |
Actually on the SRU I have one more thought.
A formerly "silent" action like the graceful restart does now emit messages:
echo "Invoking 'systemctl restart ${APACHE_
echo "Use 'systemctl status ${APACHE_
I agree this is good going forward and also good for "humans" to understand what might happen.
But on an SRU context I'm slightly more concerned people might grep output or such.
I don't have a much better idea yet, maybe pushing the messages to the log (less obvious to be seen, but less potential disruption)?
What do you think?
Bryce Harrington (bryce) wrote : | # |
Thanks for the review.
It's a fair point that in an SRU context that changes to output are effectively behavioral changes. However, I'm not sure it's worth worrying about since a) this affects just the 'graceful' command which had been a bit broken to begin with (thus the bug), and b) the new output is consistent with the output from the 'start' command. People who had been avoiding 'graceful' due to the misbehavior likely will already be dealing with the status output from the 'start' command so if due to this fix they start using graceful they may find the lack of status output weird.
That said, I do find it a bit odd that the start command outputs to stdout rather than stderr, but this seems semi-consistent with similar code. Other places in the code do emit to stderr and I don't know if this is intentional or a discrepancy. In any case, I wanted to make this a minimal fix so am keeping similar to existing code style and not fixing up inconsistencies in the rest of the script, tho I am quite tempted... Maybe someday.
Christian Ehrhardt (paelzer) wrote : | # |
Yeah, I'm ok with the argument of keeping it "the same" as start for Bionic->Groovy, thanks for the thoughts and discussion.
Bryce Harrington (bryce) wrote : | # |
Sounds good, I've tagged, pushed, and uploaded this.
As reference, that old fix for start was at https:/ /bugs.debian. org/cgi- bin/bugreport. cgi?bug= 852543 in pkg/import/2.4.25-3