Astraweb Error Compilation

[Beck38]

Day 8/216 hours:

Astra/US - 13/127 (1.1% damaged)
Astra/EU - 18/127 (2.9%) damaged
Astra/US+EU as fills - 13/127 (1.1% damaged)

EU slightly improved, first time that happened, perhaps it's because the Atlantic trans-oceanic link was slightly less loaded due to it being the weekend. We'll see if it increases error rate again in the coming days.

[Beck38]

Finally got a response to the ticket I opened on Astraweb, they are investigating. Will continue to check the test file I uploaded over a week ago, and the several nzb's going back 6 months or so to see if I see any change, once I see any change to the test file.

Wee shall see when/if things change.

Day 9/240 hours:

Astra/US - 13/127 (1.1% damaged)
Astra/EU - 55/127 (10.9%) damaged
Astra/US+EU as fills - 13/127 (1.1% damaged)

EU errors jumped, maybe because of usual Sunday PM problems; it will be interesting (as usual) to see what the next day yields.

At this time, don't see any improvements in Astra/US as of yet.

[Beck38]

Day 10/264 hours:

Astra/US - 16/127 (1.9% damaged)
Astra/EU - 23/127 (3.4% damaged)
Astra/US+EU as fills - 15/127 (1.8% damaged)

Both servers continue the downward trend; I ran a few nzb's upwards of 8+ months previous, and saw no change in the amount/number of damaged parts, so nothing seems to have changed as of yet.

[Beck38]

Day 11/288 hours:

Astra/US - 116/127 (20.0% damaged)
Astra/EU - 121/127 (33.1% damaged)
Astra/US+EU as fills - 116/127 (20.0% damaged)

Well, somethings going on; at first, I thought (and it's still somewhat possible) that there is some problem with the transmission system (I'm on Comcast, and 30-60 days ago they were having tons of problems in my area, so much so they gave everyone a $20-30 rebate two months in a row (and anyone who's dealt with them knows this is really something out of the ordinary) but I use a vpn so can route things around and head into either the US/San Jose server or the EU/Amsterdam server by several different routes, and they all came up the same, although some speed/throughput differences, but the same error rate.

Will have to see what happens tomorrow.

[Beck38]

Did more testing of my system/transmission links, and found that everything is working 100% (downloaded and tested my test files from Blocknews). There is either something serious going on with the tier one inputs to Astra/US (but would have to be mirrored by Astra/EU) which would mean either Level3/Cogentco, or something in the front-end of both Astra/US and EU. I'd tend to thing it's the front-end of both plants.

Astra is obviously doing something (at both plants); I'll continue monitoring it until things 'settle down' and give some reasonable data, but I'll continue to post daily updates as to the 'situation'.

No change on either plant of the time/date stamp on this message; both are experiencing massive error rates on my test files, will test some of the older files in a bit; the automerge function on this board is set WAY to loosely, it shouldn't be merging messages 4.5 hours apart.

[Beck38]

Things still looking pretty sorry, getting lots of connection errors, resets, you name it.

Other servers (Giganews/Blocknews) both operating without any problems at all, despite being 2-400 miles further from me (in the case of Giganews/US in LA) or 2000+ miles (in the case of Blocknews/US in Virginia); if one watches the connection speed, you can see the connections to Astra are showing intermittent speeds and such while Giganews/Blocknews are running full speed ahead without any slowdowns at all.

There may be some problems with Level3 around the SF/SJ/SC area, I can't 100% confirm this, but it could be contributing to the connection problems with Astra/US. Those who follow (or try to) these kinds of problems will remember that less than a year ago, Level3 ran into problems with it's tier1 Cisco routers when they attempted a f/w upgrade, that caused nothing but tons of problems for a couple weeks. All we need is to have two major problems beating their heads against one another.

As far as Astra is concerned, folks need to remember that the last time (2-3 years ago) that massive retention failures happened, it took at least 2-3 weeks to fix things, and it did get worse, much worse, before it started getting better. If, of course, it does get better this time; we'll just have to wait and see.

[Stehle]

Still watching this thread... and yes I hope they figure what is wrong out too.

[Beck38]

The test file I posted some 15+ days ago has been coming up >30% corrupted parts for the last week+, on both US and EU servers.

No real change on the 'older' (1-6 months) stuff that I can see (upwards of 10-20% corrupted), I haven't gone back to the really old stuff (3+ years) to see of that has changed any, but will in the next few days if I get the time.

Again, it may take upwards of a month to see any real improvements, but I'd say that without a doubt, this is the worst failure I've seen in Astraweb in 15+ years of use.

[megabyteme]

"Houston. We're getting our first look at the service module now. One whole side of the spacecraft is missing. Right by the high gain antennae a whole panel is blown out, right up. Right up to our heat shield."

[piercerseth]

"Houston. We're getting our first look at the service module now. One whole side of the spacecraft is missing. Right by the high gain antennae a whole panel is blown out, right up. Right up to our heat shield."

My god, it's full of stars.


EDIT: Astra is still borked.