lfs-timeout

[rspencer01]

I have a commit that introduces a number of large asset files into my repo, totalling 45MB. If I try to push the commit, I get the errors

$ git push
Git LFS: (0 of 4 files, 8 skipped) 0 B / 47.09 MB, 6.27 MB skipped                                                                                                                                                 
LFS: Put https://github-cloud.s3.amazonaws.com/alambic/media/147473055/ce/b2/ceb2532ac9c15e79ddab9723ee18778bf8a0db215d6fdb955e85d2c8df85e277?actor_id=2918499: read tcp 192.168.1.32:34475->52.216.1.192:443: i/o timeout
LFS: Put https://github-cloud.s3.amazonaws.com/alambic/media/147473055/c8/bd/c8bd937fef7aebca9142498d0e56b06d6e16f4d37d1c5c89834c81549414ac93?actor_id=2918499: read tcp 192.168.1.32:34476->52.216.1.192:443: i/o timeout
LFS: Put https://github-cloud.s3.amazonaws.com/alambic/media/147473055/2f/b4/2fb4a6931d72aa639cf2667bffdfbd0c8aac51dc584749ace864d79ea61def1a?actor_id=2918499: dial tcp: lookup github-cloud.s3.amazonaws.com on 127.0.1.1:53: read udp 127.0.0.1:39530->127.0.1.1:53: i/o timeout
LFS: Put https://github-cloud.s3.amazonaws.com/alambic/media/147473055/58/6e/586e95b780e556ddd0ceae04d8d30d26a104c06fa2a9a1fce4d21d0705b73bb5?actor_id=2918499: read tcp 192.168.1.32:34477->52.216.1.192:443: i/o timeout
error: failed to push some refs to 'git@github.com:rspencer01/PM4.git'
Write failed: Broken pipe

I have set my dialtimeout to 600, which I thought should be more than generous. I also note that the third timeout is local to local.

I should point out that if I watch the terminal, it seems to upload about 3MB before freezing, and then starting at 0 again. Successive running of git push doesn't result in any more progress being made (so there is no proper resumption of upload). According to the GIT_TRACE it just retries and retries.

GIT_TRACE=1 git push origin HEAD:master

[technoweenie]

There are a number of timeout settings you can try tweaking:

• lfs.activitytimeout - Sets the maximum time, in seconds, that the HTTP client will wait for the next tcp read or write. If < 1, no activity timeout is used at all. Default: 10 seconds
• lfs.tlstimeout - Timeout for TLS handshakes

Unfortunately, there is no dns timeout setting (which is related to the third timeout).

[rspencer01]

Eventually got it working with the combination of

lfs.tlstimeout = 300
lfs.activitytimeout = 60
lfs.dialtimeout = 600
lfs.concurrenttransfers = 1

Judging from what I sort of saw watching the thing, I'm guessing the culprit might have been activitytimeout. I know that my (slow) internet sometimes does hang for more than 10s. Is it worth increasing this limit? (if no, feel free to close.)

[Filimindji]

May be late, but using Azure Devops, it look like using this config settings is a good fix (that work for me) :

git config http.version HTTP/1.1

(source : https://developercommunity.visualstudio.com/t/git-lfs-push-got-413-error/867488)

[DarrenlloydGent]

May be late, but using Azure Devops, it look like using this config settings is a good fix (that work for me) :

git config http.version HTTP/1.1

(source : https://developercommunity.visualstudio.com/t/git-lfs-push-got-413-error/867488)

You are an absolute superstar! I've spent the last 24 hours wondering why this started acting up for an Unreal Project Push.

[chrisfranko]

Holy

May be late, but using Azure Devops, it look like using this config settings is a good fix (that work for me) :

git config http.version HTTP/1.1

(source : https://developercommunity.visualstudio.com/t/git-lfs-push-got-413-error/867488)

Bro... youre a god among men

[bk2204]

The infelicities of IIS and its usage with Azure DevOps are unrelated to this issue and hence off topic, but they're discussed in depth in #4211.