LLVM.org/viewvc down?

Hi,

I’ve noticed since about Thursday that llvm.org/viewvc seems to be only returning 404 error when trying to access it. Is this service going away?

Douglas Yung

It's currently disabled until the issues with llvm.org load would be
resolved. Use klaus instead.

It's currently disabled until the issues with llvm.org load would be
resolved. Use klaus instead.

If someone has time, it would make sense to make the ViewVC link point to a page saying "Not available at this time" instead of reporting "404 Not Found."

Sadly, I don't have time to look into this right now.

Regards,

John Criswell

Thanks for mentioning "klaus", I did not know about it before. Perhaps a
link to it can be added on llvm.org? (next to ViewVC)

Regards,
Johan Engelen

As an alternative to browse the sourcecode you can also use github:
https://github.com/llvm-mirror/llvm
Maybe we should look to that as well, though I don’t know how good an idea it is to link to an “unofficial llvm mirror”.

  • Matthias

Is there any eta for when viewvc will be back? I prefer to use viewvc because all of the commit emails have the SVN commit number, but not the corresponding git hashes, so it makes it easier to look it up. Or is there some easy way to map an svn commit number to the corresponding git hash?

Douglas Yung

The commit emails, including the ones in the mailing list archives,
also have links to the viewvc, all of which are now broken.

+1 for getting the viewvc back up, if at all possible.

Unfortunately, we had lot of bogus requests to viewvc (e.g. asking for
history / blame for random revision) which looked like a DoS. Most
probably we'll enable it after llvm.org will be migrated to new
hardware.

Can we just disable / limit these features? I gave up trying to use viewvc to show me revision logs a while ago because it would just sit there for ages and eventually the request would time out. It doesn’t seem to cache any of this data and it’s very expensive to get from svn.

David

How much of that are stupid (chinese) spiders? I had a lot of load
issues for the NetBSD fossil repositories until I added the following to
my nginx configuration:

  if ($http_user_agent ~* (360Spider|80legs|App3leWebKit|Baiduspider|EasouSpider)) {
    return 403;
  }

Joerg

How much of that are stupid (chinese) spiders?

You're right, the majority of them were from Chinese IPs.

I had a lot of load
issues for the NetBSD fossil repositories until I added the following to
my nginx configuration:

  if ($http_user_agent ~* (360Spider|80legs|App3leWebKit|Baiduspider|EasouSpider)) {
    return 403;
  }

I saw some other user agents as well. And they ignored robots.txt, yes.

Could you start reenabling the service with such an exclude list and
maybe slowly extend it to cover UAs from the same IP ranges produces more
than 100MB/day or so over a few days?

Joerg