Opened 11 years ago
Closed 11 years ago
#1881 closed defect (fixed)
OOS during Multiplayer Game
Reported by: | Kieran P | Owned by: | |
---|---|---|---|
Priority: | Release Blocker | Milestone: | Alpha 14 |
Component: | Core engine | Keywords: | |
Cc: | Patch: |
Attachments (20)
Change History (33)
by , 11 years ago
Attachment: | oos_dump1.tar.bz2 added |
---|
by , 11 years ago
Attachment: | oos_dump2.tar.bz2 added |
---|
by , 11 years ago
Attachment: | commands.tar.bz2 added |
---|
by , 11 years ago
Attachment: | oos_dump_diff.txt added |
---|
by , 11 years ago
Attachment: | oos_dump_yves.txt.7z added |
---|
by , 11 years ago
Attachment: | interstinglog_commands_yves.7z added |
---|
follow-up: 2 comment:1 by , 11 years ago
It's interesting that the OOS-error happened exactly between turns 3180 and 3200 where k776 deleted the CC using the dev-console. Also he was the only one who didn't get the OOS-error. I'd say deleting buildings with the dev-console caused the error and the original error is probably fixed.
What confuses me and makes me unsure about that conclusion is that I can't get the same hash as I got in my commands.txt when running it in replay mode. The hashes match until Turn 3180 but then it differs (but the hash is the same each time I try). How can the hash be the same each time but different than it was in the original game?
Btw. you should disable the profiler for replay-mode by uncommenting the savefile-function in Replay.cpp. I got a 32 GB profile.txt...
comment:2 by , 11 years ago
Replying to Yves:
What confuses me and makes me unsure about that conclusion is that I can't get the same hash as I got in my commands.txt when running it in replay mode. The hashes match until Turn 3180 but then it differs (but the hash is the same each time I try). How can the hash be the same each time but different than it was in the original game?
I got the same result (hashes in replay mode not matching those I got in the match). (I also tested with you commands.txt, but even that resulted in a different hash.
I think the issue is somehow related to the dev-console usage, so I think we can backlog this (or even close it as needsinfo).
comment:3 by , 11 years ago
Milestone: | Alpha 13 → Alpha 14 |
---|
comment:4 by , 11 years ago
A friend and I played the current release (13332-release) yesterday and kept getting OOS errors I think every time at turn 1700. I'll try to attach all info I have, please advise if you need additional files/data.
We played multiplayer on LAN, 2on2 versus two Aegis AIs on Alpine Valley.
by , 11 years ago
Attachment: | xmirakulix_host_oos_dump.txt.7z added |
---|
by , 11 years ago
Attachment: | xmirakulix_client_oos_dump.txt.7z added |
---|
by , 11 years ago
Attachment: | xmirakulix_host_commands.txt added |
---|
by , 11 years ago
Attachment: | xmirakulix_client_commands.txt added |
---|
by , 11 years ago
Attachment: | xmirakulix_oos_dump_diff.txt added |
---|
comment:7 by , 11 years ago
We tried again today with [13350]. Still reproducible with Aegis on map Alpine-generated, but we managed to play a game with qbot enemies.
comment:8 by , 11 years ago
My friend an I played numerous games on various scenarios including the Acropolis, Saharan Oases, Siwa Oasis, Oasis 1-9 and it seems just about every time, an OOS error occurs. We usually are a team against one or two Aegis Bot(s).
I run Ubuntu 12.04 LTS while she uses Windows 7. I'm attaching the oos_dump.txt file. Do you need any other logs?
by , 11 years ago
Attachment: | scythetwirler_oos_dump.txt.tar.gz added |
---|
scythetwirler's OOS dump file.
by , 11 years ago
Attachment: | scythetwirler-commands.txt.tar.gz added |
---|
follow-up: 11 comment:9 by , 11 years ago
Played another game with my laptop and my desktop (one on Windows and one on Ubuntu) with 2x Aegis. I was idle on one computer and half-playing on the other and an OOS error popped up around turn 800.
by , 11 years ago
Attachment: | scythetwirlerg2Windows_oos_dump.7z added |
---|
by , 11 years ago
Attachment: | scythetwirlerg2Windows_commands.7z added |
---|
by , 11 years ago
Attachment: | scythetwirlerg2Ubuntu_oos_dump.txt.7z added |
---|
by , 11 years ago
Attachment: | scythetwirlerg2Ubuntu_commands.txt.7z added |
---|
comment:11 by , 11 years ago
Replying to scythetwirler:
Played another game with my laptop and my desktop (one on Windows and one on Ubuntu) with 2x Aegis. I was idle on one computer and half-playing on the other and an OOS error popped up around turn 800.
Thanks for the logs. Comparing the OOS dumps from your Windows and Ubuntu machines, the only relevant difference is:
@@ -881090,7 +881090,7 @@ entities: 7357 ], "numRecentBuilders": 6, - "buildMultiplier": 0.5841906810678655, + "buildMultiplier": 0.5841906810678654, "previewEntity": 7465, "finalTemplateName": "structures/rome_barracks", "owner": 1,
which looks like a platform specific floating-point issue (of the kind reported here). I'm not sure at this point what we can do to fix that, #433 may be relevant.
comment:13 by , 11 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
OOS problems in multiplayers without AI players seem to be solved now. See #2000 for OOS problems with AI players.
Yves's logs from todays multiplayer game with k776, historicbruno, leper and 4x Aegis