On Wed, Feb 25, 2026 at 05:43:45PM +0000, Peter Maydell wrote:
On Wed, 25 Feb 2026 at 16:38, Daniel P. Berrangé <berrange@redhat.com> wrote:
On Thu, Feb 19, 2026 at 10:23:36AM +0000, Peter Maydell wrote:
Why do we care about the qemu_log output matching the error-report output? The logs are expected to be quite frequent, to only be there if you've explicitly turned them on, and to be usually directed to a log file.
Personally I never use the log-to-file facility. Instead typical use case is relying on the trace points 'log' facility as illustrated in the commit message example to dump to stderr for simple debugging, and having consistent formatting between the logs & errors is useful there.
Do you not find that the tracepoint output is often just too big to be manageable via stderr? (To take a random example, the last thing I did with tracepoint logging produced a file 196MB large for a not very long run of the guest.)
The things I'm debugging tend to be more focused on interactions around mgmt (QMP) changes, rather than guest execution, so the trace output is pretty manageable, but yeah I can see how you can easily get huge output. With regards, Daniel -- |: https://berrange.com ~~ https://hachyderm.io/@berrange :| |: https://libvirt.org ~~ https://entangle-photo.org :| |: https://pixelfed.art/berrange ~~ https://fstop138.berrange.com :|