Giganews and others change peering to Russians, massive 'Skips' result

[Beck38]

Just a heads up here, but Giganews, Usenet-News, Blocknews, and probably others EXCEPT for Astraweb, changed their peering scheme early this morning (0540 Pacific time if anyone is interested) to a Russian outfit (goblin.stu.neva.ru) as their main Peer.

This has resulted in massive 'skips' and the like in their databases.

And, this has happened before a few months ago; it took a couple days for those involved/affected to change their peering BACK to what it was. The question remains is why this change occurred... Hacked? Some ingress in Usenet? I never got a straight answer from Giganews when it happened before, I doubt I will this time.

Meanwhile, if you're trying to d/l things and getting lots of errors (and of course, the need for lots of pars to 'fix' those errors), hopefully it will be fixed in a couple of days, so hang tight and maybe it will be 'fixed' soon.

[justlooking]

Thanks for the very esoteric information. And you can be sure Vladimir Putin owns a piece of that Russian company, like he owns much of Russia.

[Beck38]

I guess it's 'esoteric' to a point, but to me it's 'basic usenet'. I watch Giganews like a hawk (great to make sure wherever things are being posted, which is usually NOT Giganews as they haven't had 'block' accounts for forever at this point and such are the real key to folks uploading, i.e., cheap and 'virtually forever', but I've always figured that if Giganews has it, most everyone does, as they've always had extensive peering every which way this side of Sunday.

Again, this happened just a couple months ago; there were a fair number of messages on many of the 'primary' binary groups complaining one way or the other. I did get 'sorted out' a couple days later, and the 'lost skips' were 'back-filled'.


UPDATE:

It appears that sometime around 3:30pm (pacific time where the GN server is) that the peering was corrected. Will wait a bit and see if they 'back-fill' over the next few hours, and then will check some of the other servers.

Asked the question of GN as to the problem, no response as of yet (don't really expect one as didn't get any the last time).

So. almost 12 hours of back-fills needed. Probably won't be able to tell if it's all been corrected for about another 12 hours or so.

[nfodiz]

Yeah this has thrown binsearch and nzbindex way out of whack... Nothing indexed for 12+ hours in the groups I checked. I'm assuming it's directly related

[Beck38]

Every time on of these 'off the rails' incidences occurs, I have to remind folks that those indexing sites, located in the Netherlands, have peering to, And they are nowhere near as 'good' as most commercial server plants. In short, on a good day they 'report' tons of skips even though taking the posted nzb (NOT the nzb generated by the indexing site) to most servers yields very little of any errors.

In short, they need to improve their feeding arrangement. But I think they figure it's 'good enough' and they are shying away from spending any more money (or programming effort) to improve things. I use them all the time to 'track' my uploading, but find that even with the propagation right at 100%, those sites report a fair number of skips and fades.

In other words, their 'reporting' has to be taken with a grain of salt. A BIG grain of salt (at times around the size of Lot's wife).

[Beck38]

It appears GN is slowly rebuilding all that they 'skipped' during their 12-hour mishap; it may be complete in another 12 hours (from now) but we'll see.

UPDATE: It appears that the backfill process has ground to a halt at present. Will see if anything actually occurs in the NEXT 12+ hours.

UPDATE2: Just got off the horn with GN, they started of claiming they have no idea what their peering is, why it was changed then changed back, etc., and any problems are due to DMCA (this on something that was in the process of being posted!). As I've pointed out to others before, if you insist on using the 'high priced spread', that doesn't absolve you of having (and paying for) an alternate feed.

FYI, Blocknews has backfilled all the parts they were missing two days ago. Perfect.

[Hypatia]

any problems are due to DMCA

LOL... It's their UNIVERSAL answer to all missing articles related problems.

Giganews has a terrible completion past 1k days I think,I don't remember exactly. And they also claimed it was DMCA related. But,hey, DMCA takedowns nowadays are done in a way that makes posts totally disappear from servers. And what we see is lots and lots of randomly missing articles, it's not a "proper" dmca pattern at all... That is unless.... their DMCA automatic system screws things up deleting articles it is not supposed to delete. Then yes, they could say that it is DMCA related :D

DMCA has become a pathetic excuse for nntp providers to not care anymore about the quality of service they provide.

[Beck38]

Basically, it's 'interesting' to me that both Astraweb/US (I don't and/or really can't follow Astra/EU that closely due to the sheer flood of headers coming from the EU, any server is hard enough to try and figure what's going on these days) and Blocknews (readnews) is running pretty good, or more than pretty good. Anyone who isn't using Blocknews as their 'second' is really missing the boat these days. Their operation really seems to be 'top notch'.

WE all know that Astra tends to go 'off the rails' occasionally, the difference between them and Giganews is that in the last x years, it seems to be few and far between, and they do jump on it at some point, and it gets repaired (eventually), whereas their 'high priced spread' competitor seems to have more 'mini skips' events and for the most part ignores them.

[MysticRiffs]

My first guess at why the "skip" occurs at all is down to time-zones, believe it or not. If they change the physical location of the server, and they accidentally travel forward in time by 6 hours... Dunno...

You're right though, this is actually hellishly interesting. :)

[Beck38]

After a few days of running their plant's peering directly from Astraweb/US, Giganews/US has been 'playing' around with their peering once again, this time grabbing stuff most directly from eweka.nl; now, they aren't exactly as 'bad' as the Russian outfit they were trying out a week ago, but as users of NZBIndex or Binsearch (both NL based engines that both use Amsterdam-based usenet providers to 'feed' their search engines), that they aren't the best by any means (lots of skips and fades).

So, there has been a lot of semi-wackiness going on with completions once again on the vaunted GN/US; haven't had the time to take a look at GN/EU, but in the past they've been very 'tight' on the same misses.

They will probably be able to 'fill-in' the skips at some point, as they always seem to several days later. Just be forewarned that if you run into the need to 'excessively repair' your GN downloads, I'd wait a few days and re-run your nzb (and don't use one 'generated' by those indexers), hopefully you've got the nzb that the poster 'posted'.