On 5/31/22 14:58, David C. Rankin wrote:
On 5/21/22 06:29, gumb wrote:
No need to specify the minor version, I say 91esr because the issue is consistent across all 91.x versions.
This version of TB is perfectly stable on my stock 15.3 box, and I have it heavily loaded up with accounts/folders/filters/storage.
I do get a KDE slow-down at start (from tracker) but that normalizes in a few minutes, and I don't use baloo. But indexers would slow down everything, not just TB.
<snip>
Will strace Tbird next to see what shows up.
Thunderbird runs two processes now: $ ps axf | grep bird 15520 ? Sl 16:48 \_ /usr/lib64/thunderbird/thunderbird-bin 15585 ? Z 0:00 | \_ [thunderbird-bin] <defunct> 15590 ? Sl 0:00 | \_ /usr/lib64/thunderbird/thunderbird-bin -contentproc -childID 1 -isForBrowser -prefsLen 1 -prefMapSize 280054 -jsInit 285716 -parentBuildID 20220502180952 -appdir /usr/lib64/thunderbird 15520 tab The second seems to be some web-browser process, likely for displaying html messages or web-pages in the message window. I have thunderbird running on a separate desktop where it is the only app running, so while strace was occurring, I was on another desktop ensuring that thunderbird received no input or focus. The problem is both processes are extremely active, a large majority of the entries are: recvmsg(4, {msg_namelen=0}, 0) = -1 EAGAIN (Resource temporarily unavailable) I took a strace of each process, and the parent process generated over 5M of output is just a few seconds. The child process a little over 100K of output, e.g. -rw-r--r-- 1 root root 5159860 May 31 15:14 20220531-151346-thunderbird-err.log -rw-r--r-- 1 root root 117214 May 31 15:56 20220531-155626-thunderbird-err.log I'm no expert in deciphering the strance output. I know what each output is telling me, but not what it means in the overall operation of thunderbird at any particular moment. But the overarching takeaway is there are a huge number of recvmsg calls where the resource is unavailable causing tbird to simply sit there spinning it's wheels. I don't know if this is normal thread blocking taking place, a deadlock, or some other cause, but it certainly looks suspicious. -- David C. Rankin, J.D.,P.E.