adblockplus.org down

[Ares2]

Looks like adblockplus.org is down, and according to the mirror logs it has been for quite some time now:

Last successful sync: Jan 21 21:05 UTC
Last successful list download: Jan 21 21:37 UTC

[runningdonut]

Is there a mirror available? Man the web is annoying without easylist

[Michael]

Ares2 mirrors all the Easy subscriptions: EasyList, for example, is hosted at https://ares2.org/easylist/easylist.txt and https://arestwo.org/easylist/easylist.txt. However, we are currently unable to update the subscriptions because the repository is located on the Adblock Plus domain. Please switch back to the normal addresses when Wladimir's server returns to ensure that Adblock Plus is aware of all possible mirrors when downloading the subscriptions.

[Michael]

Temporary plans without the Adblock Plus domain:

• Users
  • Please do not create issue reports, as these are all sent to adblockplus.org. Problems should instead be reported in the filter suggestions or the false positives forums.
  • If you need support with the Adblock Plus add-on we should to assist in most cases in the EasyList & Adblock Plus forum, although only Wladimir can resolve the more technical issues and make updates to the source code.
  • I can also provide information about other filter lists on request, including contact details, websites and subscription addresses, as I have a local copy of SubscriptionList.
• EasyList Authors
  • Please continue to commit changes to your local copy of the EasyList repository; the changes should be merged automatically when adblockplus.org returns. If using a script you should comment out the lines "hg pull" and "hg push", as these require server access.

I've also made this topic a global announcement given the importance of the Adblock Plus website and forum.

Ares2, do you know why the Adblock Plus server is down and when it is likely to return?

[runningdonut]

Thanks Michael I added a custom subscription for the mirror and kept the original one up too. Hope everything's back to normal soon again thanks for your quick help

[fanboy]

I know the feeling when the vps breaks..hopefully it comes back up soon!

[Ares2]

Ares2, do you know why the Adblock Plus server is down and when it is likely to return?

No, but the server itself is already up again (ping/traceroute), it wasn't in the morning when I posted this. The (vps) node has been reset though (different fingerprint), so Wladimir is probaby already busy getting it up again.

(Now excuse me while I'm off to "revise" the backup strategy for my own servers xD ).

[Wladimir Palant]

The server isn't up, that's only a repair VPS

I tried a Linux distribution update yesterday which didn't go too well, the server won't boot any more. That's not supposed to be an issue, after all there are backups - except that restoring the VPS from backup doesn't work. And contacting anybody at my hosting company on a weekend doesn't seem to work either. At the moment the plan is to set up the server from scratch, that should be faster than waiting until on Monday somebody can look at why restoring the backups isn't working (and assume that the problems isn't with the backups themselves). I have my private backups, only thing missing are the issue reports which I am downloading right now via repair system (getting 4 GB out of it isn't exactly fast). Once I get them I should have all the current data, nothing will be lost.

[Michael]

That's not supposed to be an issue, after all there are backups - except that restoring the VPS from backup doesn't work.

Aren't your hosting company meant to check that the backups work? There is no point in the functionality if the system cannot be restored from the data.

[bee]

Hi!!!!!!!!!!!!
YEAH!!!!!!!!!!! private backups are very smart!!!!!!!!!!!!!!! I also keep the original copy of my website, saved in the hard disk of my computer!!!!!!!!!!!!
It's very unfortunate when after an update LINUX quits to work but it's always a fault of the distribution maintainers
An update a day, keeps the kernel panics away!!!!!!!!!!!!!!!!!!!!!!!!!!!
For some distro, i know, it's safer to update everyday rather than once in a month!!!!!! ArchLinux

bye!!!!!!!!!!!!!!

[Wladimir Palant]

@Michael: Of course they are supposed to. Which doesn't mean that they do - or they assume that if it works for one VPS it will work for the rest of them as well. Unfortunately, I have no idea what exactly is broken and whether the backups that are being displayed are valid.

I will try to reach the support one last time now, if it doesn't work I will set up the system from scratch.

[Wladimir Palant]

Supposedly, the backups won't work on a system that isn't runnable - so I have to reset the system first before the backup can be installed. I triggered the reset an hour ago, still waiting for it to actually happen.

[Ares2]

I triggered the reset an hour ago, still waiting for it to actually happen.

That doesn't sound good, if it's not automated, you might even have to wait until Monday?

[Wladimir Palant]

It is automated. I have the suspicion that the server reset is hanging for the same reason the backup wasn't working.

I now managed to start Apache in the repair system to show a message at least. However, the resources of the repair system are extremely limited, it cannot handle the connections (even without having the SSL port open). http://easylist-downloads.adblockplus.org/easylist.txt works somewhat however...

[Ares2]

Well, then I would say temp-redirect as much as possible to my server to not pile up your very own DDOS army (aka failed subscription download retries)?

[Wladimir Palant]

I cannot redirect anything without accepting SSL connections.

[Ares2]

I know, but I thought you could get that working as well. Connection limiting should be easy with iptables.

[Wladimir Palant]

No, the system only has Apache 2.2.3 without mod_ssl (prefork version meaning that it is bound to hit the process count limit). Installing Nginx is impossible because the file system is write-protected and the RPM is not relocatable. I cannot compile it either, no C compiler on the system.

[Ares2]

No, the system only has Apache 2.2.3 without mod_ssl (prefork version meaning that it is bound to hit the process count limit). Installing Nginx is impossible because the file system is write-protected and the RPM is not relocatable. I cannot compile it either, no C compiler on the system.

Oh.... OK, sorry. I thought you had full access to your disks.

Isn't that what a rescue system is all about? Copy over $important_binary which you decided to remove it with your genious scripts? Fix the occasional stupid ssh/network config error you locked yourself out with? (I know I have used mine for that more than once already ).

[Wladimir Palant]

Ok, I was being stupid - I already have nginx on my original system which can be started and works well. However, for SSL it is trying to use the library in the repair system which is very outdated and fails of course.

[Ares2]

Can't you just chroot it to /mount/mydisk and put your config together with the commandline options?

[Wladimir Palant]

Don't know, maybe. For now I brought up the static part of the site which is at least something.

[Wladimir Palant]

It is running chroot'ed now - but without SSL. Opening the SSL port immediately causes the VM to overload, even with nginx.

[Ares2]

What about limit new 443 connections via iptables and only putting the easylist-downloads subdomain on ssl? I have never tested it, but if 1 of the 2 connections Firefox creates, gets through and replies with a 302, it should be enough, right?

Code: Select all

iptables -A INPUT -p tcp --dport 443 -m conntrack --ctstate NEW --tcp-flags SYN,ACK,FIN,RST SYN -m limit --limit 1/second --limit-burst 4 -j ACCEPT

iptables -A INPUT -p tcp --dport 443 -m conntrack --ctstate NEW -j DROP

If iptables is old, state instead of conntrack:

Code: Select all

-m state --state NEW

List rules:

Code: Select all

iptables -L -vn

Clear added rules:

Code: Select all

iptables -F

[Wladimir Palant]

I found the reason why the system wasn't booting, now it is up again. FastCGI servers aren't starting up but that's a minor issue.

Edit: FastCGI servers are up, a utility moved to a different directory in the new distribution version. Now it is only the database...
Edit2: Database is up as well.

[Wladimir Palant]

@Ares2: I've been playing around with iptables a while ago - in the VM you have to do with the modules that are already loaded into the kernel and my webhoster didn't load too many. Most of the interesting iptables modules weren't loaded and I'm pretty certain that conntrack is one of them.

[Ares2]

I found the reason why the system wasn't booting, now it is up again. FastCGI servers aren't starting up but that's a minor issue.

Edit: FastCGI servers are up, a utility moved to a different directory in the new distribution version. Now it is only the database...
Edit2: Database is up as well.

Wohooooooo. hg, rsync?

in the VM you have to do with the modules that are already loaded into the kernel and my webhoster didn't load too many. Most of the interesting iptables modules weren't loaded and I'm pretty certain that conntrack is one of them.

Ah, right. That (and all the strange limits) is why you never want to get an OpenVZ/Virtuozzo VPS but a "real" one (Xen, KVM etc).

[Michael]

I can confirm that the EasyList website may be administrated correctly, including the blog, although would reiterate the request that Mercurial access is restored as soon as possible.

[Wladimir Palant]

For some reason SSH sessions with hg and rsync are closed immediately, the log doesn't even list them. I doubt that I can look into this before tomorrow, already too much of my weekend spent on making the server work

PS: Changed SSH log level - session appears to be rejected by PAM. Now I only need to figure out why...

[Wladimir Palant]

Turns out, distribution update disabled login for users hg and rsync, for whatever reason. Fixed, everything should work now.