Hi,

when I try to push to my fossil repo I get the following:

---
$ push -v
Push to https://my.domain.org/path/to/fossil/repo
                Bytes      Cards  Artifacts     Deltas
Sent:           39292        559          0          0
Received:       35576        502          0          0
waiting for server...
---

Then nothing happens and after some time the process exits.

In my web server log I do see this POST request (anonymized):

---
my.domain.org x.x.x.x - - [31/Jul/2022:11:27:09 +0200] "POST /path/to/fossil/repo HTTP/1.0" 200 0
---          

The fossil.log remains empty.

Cloning/pulling does work however.

I did check the filesystem ownership/permissions settings on the server and as well as the fossil user permissions and couldn't really find an error.

How could I further debug this?
I assume it would be good to somehow get more verbose output to find out why the push really fails.


Thank you very much.

> Does "sync" hang in the same way?

Yes it does.

> What version of fossil are you running?

---
$ fossil version
This is fossil version 2.18 [84f25d7eb1] 2022-02-23 13:22:20 UTC
---

Same for server and client.

> What OSes (client and server)?

Both are OpenBSD.

Update on the issue:

I noticed that the data to be synced contained a larger binary file (~50MB).
Checking in only the other smaller files the sync worked without a problem.
I then checked-in the binary file as unversioned and did

---
$ fossil sync -v -u --httptrace
...
                Bytes      Cards  Artifacts     Deltas
Sent:            3175         43          0          0
Received:        3715         54          0          0
Unversioned-file sent: path/to/large/file.bin
waiting for server...
---

I do get a file `http-request-2.txt` which contains the binary data of the file.
The file `http-reply-2.txt` remains empty.

Running `slowcgi` in the foreground with verbose mode eventually gives:

---
...
slowcgi: fork: /cgi-bin/scm
slowcgi: caught exit of unknown child 90939
---

Since I followed fossils' openbsd fastcgi setup instructions [1] I doubt that it is a "max request body" problem having the following in my httpd.conf:

---
connection max request body 104857600
---

I suppose then this may also be a none-fossil issue. 

It still would be nice to produce/find meaningful output that describes the origin of the problem. I will continue to debug this and hopefully find a solution.

Reading about the fossil sync protocol [2] is interesting.

[1] https://fossil-scm.org/home/doc/trunk/www/server/openbsd/fastcgi.md
[2] https://fossil-scm.org/home/doc/trunk/www/sync.wiki

I believe to have found the answer to the problem:

The default timeout of openbsd slowcgi is 120s [1]. After that the process is "cleaned". I recompiled slowcgi with a 600s timeout. Now the request is going through.

So fossil works just fine. 

Thank you very much!

[1] e.g. https://github.com/openbsd/src/blob/master/usr.sbin/slowcgi/slowcgi.c#L41