I currently have 3 WUs (each on a different machine) that have been running for 6, 7, and 9 hours. Is this normal and should I let them continue to run? These are all on Linux with app version 2.26

If it helps,last entry in solver_stdout.txt file:

[2026-09-30 07:22:35] cost funnel d244 6/6/6/2/1/1 d248 0/0/0/0/0/0 d250 0/0/0/0/0/0 d252 0/0/0/0/0/0 (slack 0/1/2/3/4/5)
[2026-09-30 07:22:35] cheapest-completion ring mismatches: 0:129 2:4
[2026-09-30 07:22:35] ladder rung 1 (rewind 236): started 133 finished 133 inflight 0 capped 0 nodes 178.0M (1338670/call) gap-to-budget +2:9 +3:50 +4:74
[2026-09-30 07:22:35] ladder rung 2 (rewind 230): started 59 finished 59 inflight 0 capped 0 nodes 1404.6M (23807237/call) gap-to-budget +2:13 +3:46
[2026-09-30 07:22:35] ladder rung 3 (rewind 224): started 13 finished 12 inflight 1 capped 0 nodes 43088.7M (3590726620/call) gap-to-budget +2:12 progress 305M/64000M (0.5%)
[2026-09-30 07:22:35] frontier-dedup: termCache hit 0 miss 13 (hit% 0.00) | chain-enqueue new 0 dup 0 (dup% 0.00)
[2026-09-30 07:22:35] sigma-MAIN(OFF): breakCands 0 wouldCut 0 (0.000%) cut 0 leafBugs 0
[2026-09-30 07:22:35] sigma-TAIL(ON): nodes 1249452978396 wouldCut 1206166142028 (96.536%) cut 1206166142028 leafBugs 0

Thanks for the detailed report — the log helps. Yes, please let them run. What you're seeing is normal solver work: each task explores a slice of the search, and a few slices go much deeper than average. On your machines most tasks finish in about 1 to 2 hours, but a small share take several hours; the longest one completed so far on your hosts took a little over 7 hours. Importantly, the current Linux version can't resume a stopped task, so aborting now would throw away those hours. Resume support is being built for the next version.

Interestingly, I suspended ALL tasks on ALL machines (7) yesterday for 6 hours and then resumed ALL tasks. They all seemed to complete without errors. At least they didn't end up in the error category.

Good to hear, thanks for testing that. From now on it's even safer: version 2.31, released today, adds proper resume, so even a full restart of BOINC or the machine no longer sends a task back to zero. Your Linux hosts will switch to it for new tasks automatically.

2.31 looks like a big improvement. All 2.31 WUs are reporting frequent checkpoints and running faster.
Credit is still 32. I thought I saw it was going up to 150.
https://boinc.eternityathome.org/eternity/forum_thread.php?id=4&postid=64#64
"...new work is sized to about 30 minutes each at 150 credits."

---

there are still some issues

- 32 credits but could be some older WUs?
- running in background while suspended
- 0 elapsed time after restart (BOINC wrapper issue, it is the same on SRBase) but not a big deal because of fixed credits, checkpoints are logged in stderr.txt

My 2.31 WUs don't behave like that.

Might want to check this task as it has been running for over 24 hours on a system that normally is under 2 hours run time.

https://boinc.eternityathome.org/eternity/result.php?result_name=eternity_cf_xp_explore-f2-r2_s258_r2003769517_2003770030_0

Thanks

In reply to rebirther's message of 30 Sep 2026:

there are still some issues

- 32 credits but could be some older WUs?
- running in background while suspended
- 0 elapsed time after restart (BOINC wrapper issue, it is the same on SRBase) but not a big deal because of fixed credits, checkpoints are logged in stderr.txt

I can tell you that all these issues are fixed.

I had the exact same moment right now ..even on 2.33 WUs ^^

Thx for the manual solution to suspend and "restart" them to get them finished xD

---

---------- 24/7 Crunching since 2011 -----------
DSKAG Austria: http://www.dskag.at

Merci pour les retours. cellarnoise2 : cette tâche tournait sur l'ancienne version 2.29 et a atteint la limite de durée de BOINC au bout d'environ 42 heures, d'où son arrêt. Désolé pour le temps perdu. Aurum : le plafond de crédit est corrigé, les résultats reçoivent maintenant 150 ou 300 selon la taille de la tâche. dskagcommunity : pouvez-vous me donner le nom d'une tâche 2.33 qui est restée bloquée, et ce que vous avez observé ? J'aimerais la vérifier directement.