Astraweb/US, Possible Changes To Front-End Plant

[Beck38]

Over the last (insert months/years here), Astraweb/US (maybe Eu as well...) has had front-end problems with their connection processor; resulting in dropping of connections, excessive time-outs, that kind of stuff (this is all on SSL).

The last couple of days, I've noticed that when the front-end drops or otherwise has problems with a connection, instead of dropping and re-negotiating that single connection, several (usually three) connections are dropped and re-negotiated. So far (will of course be watching) this seems to have eliminated what I'll call 'tripping', where one connection 'trips out', re-negotiates, then trips yet again (sometimes over and over).

Now, it seems to be re-negotiating a 'triple' (or triplet) of connections, and that seems to perhaps 'calm' (or resets) the multiple connections one may be using.

Again, I'll see if it continues to perform better in the long run, perhaps someone there actually took my advice and started looking hard at their logs and such, and decided to do some proactive maintenance.

One CAN hope....!?!

[Hypatia]

Has anyone been having connection timeouts recently?

Specifically during the upload process... it drives me mad.. uploading something larger than 15G has become unbearable..
sometimes trying different server ip address helps..

[sandman_1]

Newsbin Pro usually shows for me some connection issues with Astraweb. However I don't think much about if my line is still maxed out. I hope they are doing some much needed maintenance.

[Beck38]

Has anyone been having connection timeouts recently?

Specifically during the upload process... it drives me mad.. uploading something larger than 15G has become unbearable..
sometimes trying different server ip address helps..

If one is using anything other than JBinUP to post these days, especially with a fast connection (>1Mb/s) on large (>10MB) rar posts, they need their head examined; I did use PowerPost a ways back (still do on small quick posts) but quickly found out that on large postings it wasn't up to the task.

I run my line pretty near at 100% cycle either posting or downloading, and I get maybe one error in 60+ days with JBinUp that has to be manually 'restarted', if that. All on Astraweb. But I am a bit (some might say more than a bit) watching like a hawk all the time. If my plans to move in the next year or two and get on fiber at 5-10 times my current speed (I'll have to upgrade my router plant at a minimum to handle the speed) then there's no way older s/w will handle it.

I really can't say enough about JBinUp, run 24/7, and even on a plant with the 'more than occasional' hiccup (Mr. Astraweb), it goes through all these kinds of hiccups pretty well, Astra does have excellent propagation and that's the key.

[Hypatia]

well..ive been using jbinup for a long time now =)

it seems to me the problem might happen when jbinup tries to do its header check while still uploading
at least it looks that way

pS it seems to be working OK again..

[sandman_1]

Having some connection issues with Astraweb this morning. Speed shoots up to max then just drops to almost nothing after grabbing a few articles. The connections in Newsbin show all 3 of their servers doing the same thing at the same time. IDK wtf they are doing...

[Beck38]

It would really be nice if folks reporting something 'going on' with a particular plant would properly identify WHICH plant they are talking about.

For instance, simply saying 'Astraweb' when there are two different plants on opposite sides of the planet tell me nothing.

Be that as it may, Astraweb/US did take a 'burp' around 3am local San Jose time this morning. From the logs it recovered quickly and has been running fine since.

Although Filesharing Talk is acting strage today.... very slow on the uptake and then seems to have problems posting.

[sandman_1]

It would really be nice if folks reporting something 'going on' with a particular plant would properly identify WHICH plant they are talking about.

For instance, simply saying 'Astraweb' when there are two different plants on opposite sides of the planet tell me nothing.

I thought I made that abundantly clear in my post:

he connections in Newsbin show all 3 of their servers doing the same thing at the same time.

All 3 of their SSL servers (really 2 but they list 3) that is to clarify even further.

[sandman_1]

Delete, double post.

[Beck38]

Nope, still don't.

Astraweb has TWO plants, one in the Netherlands (Amsterdam) and one in the US (San Jose/Santa Clara, CA). The SSL connections are simply through an additional 'front end' at each plant, and the IP address some users use do provide a 'non-specific' set of addresses that will (supposedly) auto-route to the nearest (internet hop count/delay) plant.

I think everybody is, or should be, of the opinion NOT to use those addresses and instead use the direct ones to whichever of the two plants is 'closer' and provides the most stable connection, and not rely on whatever intelligence (or lack thereof) of some auto-route scheme. That will help in the long run.

And yes, it looks like the server for Filesharing Talk is getting a bit weird today with the double postings...