Astrraweb/US going 'off the rails' again with massive skips/incompletes

[Beck38]

Started to notice this about 12 hours ago; I had just finished getting something over 2 years old (no errors whatsoever) and did a quick check on something a bit less than a month old, and saw massive incompletes (it was 100% a couple weeks ago).

Will be checking the EU server shortly, but in going back 2-3 months it appears that the US server is 'falling on it's face' once again. A ticket has been launched at their site, and I'll continue to check/re-check things to figure out where the 'gap' is in their US plant.

So if you're trying to get something off AstraUS in the next few days, be advised. Hopefully, the EU server is running okay, but will take a few hours to confirm that (and anyone who uses US as their primary should have the EU set up as their first secondary anyway).

UPDATE: The EU server appears to be fine; again, if you're using the US as primary and EU as secondary, you'll be grabbing a whole bunch of parts off the EU. IF it's the other way around (EU as primary US as secondary) good luck getting anything good out of the US.

UPDATE2: In going more in-depth, things are disintegrating more, the number of 'skips' is increasing, almost doubling over the last 24 hours

UPDATE3: It looks like the affected 'region' is about from some 3-4 weeks ago to around 6 months previous (~1 Oct 2013); I'm doing some 'deep analysis' the next over the next 12 hours or so, should be able to nail it at that time.

[freelance86]

ive noticed for the past 12 hours that im getting a shitload of 502 errors (1 hour good, 1 hour bad, 1 hour good, 1 hour bad etc etc), cant really tell if its packet loss through my isp or astraweb so ill just blame astraweb for now... everything im getting is complete so i dont know if they are related or not, however im grabbing stuff thats 2-3 years old.

edit: using astraweb us ssl.

[bikertyke]

I moved from Astra nearly a year ago. Service became very poor. I moved to Usenetnow and never looked back. Hardly any bad blocks on anything.

[007av]

I'll join the party, just purchased a 3 month subscription b/c it was a good deal... regretting it now.

[Beck38]

UPDATE4:

All confirmed, the 'gap' is from around the 10th of March back to around mid August 2013; back further, it all looks good.

It has been reported, then they (Astra) closed my ticket; I'll reopen it.

The Astra/US plant in San Jose is literally a hop and a skip from me internet wise; those plants on the east coast have to crawl the bits across the continent, and a s-l-o-w by comparison to both Astra/US and Giga/US (both in California), and of course the EU plants (mostly in Amsterdam) are twice as far and go across the Atlantic on submarine cables.

So for those in the western US, there really isn't any good alternative (for a primary plant). Hopefully, they'll fix whats wrong (they have in the past) and things will get back to 'normal'.

[arnesr]

I noticed the issue as well. posts look complete on nzbindex.nl but lots broken parts.

[Beck38]

NZBIndex has historically been at least minimally (to worse upon occasion) with getting all the parts of things, which is why I always tell folks to use the POSTED nzb unless it wasn't (by the original poster) and just be advised that you'll most probably have to do some decent repair job (hopefully lots of pars2's).

But they have been pretty bad over the last few weeks. They appear to 'follow' (main peering) off Giganews, which has been doing pretty bad with things posted from Astra, again, the past couple of weeks. If one looks at the peering (at Giganews) you'd see that they 'normally' directly peer straight from Astra, but occasionally change it up (for whatever reason) and that's when things get wobbly there; however, they usually 'catch up' with the postings in 2-4 days or so, but the damage to folks like NZBIndex has been done.

We all tend to think things work pretty much 'instantaneous' with usenet, but a fair amount of time it really doesn't. I can remember back when the EU servers were DAYS behind the US (and vice versa, depending on where things were posted) and the largest trans-Atlantic 'pipe' was barely 500Mb/s (coaxial, pre-fiber days).

We're spoiled, but that still doesn't mean that folks should be lax watching things....

[freelance86]

I noticed the issue as well. posts look complete on nzbindex.nl but lots broken parts.

as with most indexers, they all "look" complete until you try and download them but for whatever reason the part is missing. could be the server they use for the index is different to the server you use to download so propagation is bad.

something was up last monday as the majority of tv shows posted were missing a lot of parts, most were repairable but 1 i had to download a different post and mix and match the files and then repair it. fun times...

edit: no service is perfect and even the overpriced giganews has the same issues, i just wish astraweb could get with the times and allow more then 20 connections to combat the 502 errors i get on occasions.

[justlooking]

I only hope this can be fixed by April 6th, because as Beck38 knows so well, that's Wrestlemania day. ;)

"WrestleMania XXX will be the 30th annual WrestleMania professional wrestling pay-per-view (PPV) event produced by WWE to be held, on April 6, 2014, at the Mercedes-Benz Superdome in New Orleans, Louisiana. The event will be the first WrestleMania to be held in the state of Louisiana.
WrestleMania XXX will feature professional wrestling matches involving different wrestlers from pre-existing scripted feuds, plots, and storylines that play out on WWE's television programs. Wrestlers portray heroes or villains as they follow a series of events that build tension, and culminate in a wrestling match or series of matches."

[piercerseth]

I only hope this can be fixed by April 6th, because as Beck38 knows so well, that's Wrestlemania day. ;)

:lol: Thrones premieres that night. Wouldn't be surprised if the viewers kill HBO GO again, and then watch usenet come grinding to a halt a few minutes after it posts there. Figure about 03:00 GMT.

[Beck38]

Astra got back to me (they're confirming/working on it) and it appears from the slowdowns today that some beating on the plant has begun.

So far, I haven't seen any problems with outgoing or incoming propagation, but I have seen an overall slowdown in outgoing transfers (better known as 'leeching') although there may be differences as to whether the files are old/new/whatever.

[Beck38]

UPDATE: It appears as though the latest problem is being repaired, I noted that where the 'gap' started (around 5-10 March) it has been pretty well cleared up (completion around 99+%) and am currently re-running checks going back a month or so (to around 1Feb).

If that checks out good, I'll re-run checks back the 6+ months and verify that 'all is good again' on Astra/US. It will take some time (a good 12 hours from now should do it) so hopefully by then things will be okay.

During all this, the EU server seemed to be just fine, so again, US folks that put the EU server as a 'secondary' in their d/l'er probably didn't seem much if anything unless a bit slower as the bits had to 'cross the pond' a bit more than usual.

Yet Another Update:
Looking good going back 6 months, doing a wide check right now, will take another 12 or so hours, but it's looking really good so far, everything at 98-99+% complete if NZBcc can be fully believed.

This is the third or fourth major 'gap' I've found in the last couple of years (I must admit that they've jumped on them pretty quickly and repaired them all in a couple weeks), I wondered (to them) if Astra's program (if any) to catch these anomalies is 'up to the task' if if takes some user to find and figure them out, maybe they need to 're-visit' their automated (or manual) system to be more pro-active in fixing things. (this pre-supposes there is any program in place...!)

P.S. For those who were seeing slowdowns and such over the last week plus. I notice after this repair, things have been VERY speedy, even going back a couple of years or more.

[Beck38]

Completed a review going back some 4 years (to 2010) and things look great (worst case, 99.4% complete). Please remember, these are on files that were all posted using JBinUP in full verification mode, and were re-checked a day or two after posting with NZBcc reporting 100% on BOTH Astra/US (where they were posted) and on Astra/EU, AND re-checked some 30 days later and still reporting at 99% or better.

'Just fer fun' I'm running a REALLY deep verification back to 'day zero' of Astra retention (currently mid August 2008) although obviously JBinUp and NZBcc didn't exist at the time, I utilized other tools to verify posting and retention at the time, and it will be interesting, as it always is, to see how things that old are 'fairing' in the storage plant.

'Looking Back' showed that things are pretty much 96%+ back to around Oct08 and begins to pretty much drop off before that, by the time it gets into Aug08 it's almost completely gone (gets below 10% recoverable). But that's really pretty good, I would say that anything back to around 1 Jan 09 is 'good to go', now if there were only decent indexers that could go that far back...!

Anyway, things are looking good, until I note things going 'off the rails' the next time, and hopefully perhaps they (Astra) can come up with a good scanning program (human or automated) that will catch problems before they become problems.

[myself12345]

Its still all messed up all my headers are incomplete and NZB from up-loaders who are posting with astraweb are garbage and they propagating to other servers like this so it effects everyone.
This really sucks Ive never seen it this bad

I emailed them also and got the same BS reply that they are working on it.

edit

Seems to be working again...but for how long.

[Beck38]

Propagation is at the top of my list of checking things, including several major plants (Giganews, Blocknews, Usenet-News, etc), and over the years, propagation OUT of Astra/US has been virtually perfect.

Not to say the INGRESS at some of those plants (Giganews in particular) don't go off the rails for a day or two at times, then take another couple of days to 'catch up'. In the meantime during those events, the other major plants have absolutely no problem with receiving and keeping up with the Astra/US feed. In looking at how Giga is being run during these events, they have invariably changed their peering (out of the blue) so that instead of getting a 'direct' feed from Astraweb they change to a feed out of (insert strange usenet plant out of the other side of the planet, like RUSSIA).

Then, like I said, a couple days later (out of the blue) they change back, and within a couple days 'back fill' all the parts they missed while playing buds with x.

Everyone who does any large scale work with Astra/US knows of the Sunday PM slowdowns; lately, it tends to come and go, and be a bit less time wise (2-3 hours) instead of what it was a year ago (6+ hours usually). Simply avoid that chunk of time, and you're probably okay.

But like I reported above, it looks like the entire 2000+ days of retention are good to go, and I see no slowdowns at present (up to my present ISP bandwidth, of course).

[myself12345]

well i download mainly headers from astraweb and ton of incomplete today and those post are still incomplete ATM so i dont see how retention are good
Like i said astra is still having issues


This was pretty much every server ATM things seem OK but im sure it will go to shit again.

http://i.imgur.com/wovrihm.png

[DaDD]

Seems to happen once a week lately. Not on astraweb but groups on my providers feed is filled with their post

[Beck38]

Seems to happen once a week lately. Not on astraweb but groups on my providers feed is filled with their post

What 'provider'? Your ISP usenet server or what... identify.

[TheRoo]

Really slow speeds for me and lots of incomplete this week.

[vigor5000]

I downloaded a nzb earlier today and it completed(about 1 year old post) 30 mins later I tried same one again from different nzb sites too and it wont download now.I am testing few ngp(5-10)too and have few block accounts.anyone can tell me wht its not working and what is that internode server?

[Beck38]

I downloaded a nzb earlier today and it completed(about 1 year old post) 30 mins later I tried same one again from different nzb sites too and it wont download now.I am testing few ngp(5-10)too and have few block accounts.anyone can tell me wht its not working and what is that internode server?

Don't quite understand what you did or doing..? What do you mean by 'a different nzb site'? And what you mean by 'internode server'?

[vigor5000]

Lol my bad, I should have worded it better.What I meant was I downloaded a file today using astraweb + back and block accounts, then 30-60mins later I kept trying to download the same file again but it would not download.I thought maybe I didn't grab the right nzb so I tried different ones from different indexes and now I cant download the file anymore.This "news.internode.on.net" server what is it for ? I managed to use it for liek 500mb with my astraweb account but now it doesnt work really.

[Beck38]

Not a PPV pro-wrestling capture...? Sounds like a DMCA but if on Astra should you should/would have gotten a message (I would think).

Don't have any idea as to the 'news.internode.on.net'. Just unknown, are you using NZBcc by any chance?

[vigor5000]

Not a PPV pro-wrestling capture...? Sounds like a DMCA but if on Astra should you should/would have gotten a message (I would think).

Don't have any idea as to the 'news.internode.on.net'. Just unknown, are you using NZBcc by any chance?

I found out later that is a server for Internode ISP.I tried NZBcc it shows it is missing most articles and then it always stops checks around 80-90% is that normal?it does not give me completion %.I also get a SSL negotiation failed with one of my servers in NZBcc,not a Astraweb server.

[Beck38]

NZBcc has kinda stopped development, so there are some wacky things with it (the original developer is trying to sell the thing).

Giganews for instance seems to go really s l o w, they may have tweaked something in there front end connections to make it slow when it sees someone trying to verify, who knows.

My personal opinion (take it for what it's worth on mr. internet) is with all the obfuscated and encrypted stuff filling up usenet these days, even without all the 'quick downloader' schemes and private websites hosting the 'keys', one would think that they'd generally take the hint (and perhaps they have except for the occasional takedowns of really obvious stuff.

[Ellios]

I just picked a film off Astraweb 902.7 days old, for me at least that's a record!

[Beck38]

Astra 'claims' they have 2057 (which calculates to 16 Aug 2008) but in my NZBcc runs back to what I've posted (known good at the time and for several weeks/months after) it's really more like Oct 08.

I generally work on 1 Jan 09 back date 'for sure', and will typically get something way back into the 1300+ days and more back. The problem, as I've noted here upon occasion, is that the 'free and open' indexers (NZBIndex/Binsearch) only go back 1000 days or so.

All this 'stuff' and one really can't get it or even know if it's there, if one didn't grab the nzb at the time, is 'hidden'.

[Kayslay]

just switched to usenetnow from astraweb. thanks

[Beck38]

As I'm sure you've read here, Astra 'does' 'have it's moments', so to speak (BAD).

Basically just avoid Sunday PM (San Jose time for Astra/US, you didn't specify which) and the euro folks know the EU server 'bumps and grinds' pretty well.

Maybe once per week I note where the US plant grinds to a halt, other than the Sunday PM; early Tuesday morning they came to a halt for about 2 hours or so. I was 'asleep' (it happened around 2am pacific time) but my logs picked it up, plus things were off by about those 2 hours.

Otherwise, works fine by me, and the price is right for unlimited. Dodging the SundayPM and then the slowdowns from my cable ISP (later afternoon/early evening for transferring from the plant to me), and things work pretty good.

[PastTense]

The problem, as I've noted here upon occasion, is that the 'free and open' indexers (NZBIndex/Binsearch) only go back 1000 days or so.


NzbClub is free and open and goes back 2,000 days or so:
https://www.nzbclub.com/

[Beck38]

Once again, do you folks actually READ the faq on that site? Apparently not. Or do some basic 'testing' of just how 'deep' their indexing actually runs...?

The '2000 days' is only on SELECTED newsgroups, most of them foreign (non-English) and certainly not any of the heavy Bluray or DVD groups.

[pokitisme]

is there a ue server addy for newshosting i could go thru the steps to find out but im being lazy and when i tried to get the addy for giga when i was using them they gave me a bunch of shit opinions and general crap

[Beck38]

It appears both Giganews and Newshosting (Giganews I know absolutely for sure) use a 'single url' auto-routing type system that looks at YOUR address and FUGGERSOUT the best server (right!) for you.

So in both cases, you don't have a 'choice'. That may be a real hindrance in certain cases. They of course think that it makes things 'simpler', as they feel most of their customers can't walk and chew gum at the same time.

[smgoverno]

just switched to usenetnow from astraweb. thanks

So after a few days how do you rate usenetnow compared with astraweb?