[Xapian-tickets] [Xapian] #814: 1.4.18: test suite i failing and runs forever
Xapian
nobody at xapian.org
Wed Sep 15 21:51:40 BST 2021
#814: 1.4.18: test suite i failing and runs forever
----------------------------+-------------------------------
Reporter: Tomasz Kloczko | Owner: Olly Betts
Type: defect | Status: new
Priority: normal | Milestone:
Component: Other | Version:
Severity: normal | Resolution:
Keywords: | Blocked By:
Blocking: | Operating System: All
----------------------------+-------------------------------
Comment (by Olly Betts):
> Only when I run it with -j1 I was able to see below:
Prsumably the lack of output with `-j48` in effect is due to `make -O`
which AIUI buffers all output for each job until it is complete, so a job
that doesn't complete will presumably show no output.
What distro and CPU architecture is this on?
How is configure run? You can see with `./config.status --config` if
that's easiest.
And is the packaging applying any patches?
It's strange for `zerodocid1` to be problematic as it is very simple - it
searches a single document database for a term that should match and
checks that there's exactly one returned result and it has a non-zero
docid. I'm wondering if `make -O` is causing some extra buffering even
with `-j1` and actually things stall a little later.
Either way, finding out where it's getting stuck would be useful.
Can you see if rerunning without valgrind makes a difference? You can do
that with:
```
make check VALGRIND=
```
If it still hangs, find the pid of the `apitest` process and attach `gdb`
to it (e.g. `gdb --pid=12345`) then `bt` inside `gdb` to get a backtrace.
If it only happens under valgrind, I think you can't usefully just
attached `gdb` as then you're debugging valgrind itself, and instead we'd
need to use valgrind options like `--vgdb` but let's only worry about that
if we have to.
--
Ticket URL: <https://trac.xapian.org/ticket/814#comment:1>
Xapian <https://xapian.org/>
Xapian
More information about the Xapian-tickets
mailing list