[4+ EH] Slush Pool (slushpool.com); Overt AsicBoost; World First Mining Pool

[allinvain]

hmm..slush must be fscking around with the server cause it keeps on going "down for maintenance"...

wonder if this is the planned long polling upgrade...

[andyb]

hmm..slush must be fscking around with the server cause it keeps on going "down for maintenance"...

wonder if this is the planned long polling upgrade...

I don't think so... this is the 3rd day when the server drops at around this same hour. It will pick up in half an hour or so. slush said he will ask the provider what's going on. Hope he finds out.

[allinvain]

hmm..slush must be fscking around with the server cause it keeps on going "down for maintenance"...

wonder if this is the planned long polling upgrade...

I don't think so... this is the 3rd day when the server drops at around this same hour. It will pick up in half an hour or so. slush said he will ask the provider what's going on. Hope he finds out.

I think you're right cause now the server is responding as expected. What an odd issue. Could all the miner be overloading his provider's backbone links? But that would not explain why this is happening regularly at a set time.

[slush]

I don't think so... this is the 3rd day when the server drops at around this same hour. It will pick up in half an hour or so. slush said he will ask the provider what's going on. Hope he finds out.

I was watching today's outage whole time, with another sysadmin. We didn't found anything wrong with application, but internal routing looks corrupted. It is crazy that it is always at the same time and after some time it recover itself. Linode guys are saying that everything works on their side, but I'm now absolutely sure that those issues are related to infrastructure. So I'm forced to change provider, which unfortunately will take some time.

[nzbtc]

So what time does it usually fix itself?

[cablepair]

yeah thats weird bro. I have been watching it too every day around this time.

Are you sure the server does not do any maintenance tasks at this time of day?

Or perhaps your uplink does?

very bizarre.

[cablepair]

So what time does it usually fix itself?

it usually starts sometime after 10:00 UTC and then recovers before 12:00 UTC
it varies with ups and downs in between that

[urtur]

It may be helpful that while the server malfunctions refreshing the web page results often: "Error 502 - bad gateway"

Edit: or "Error 504 - gateway timeout."

[slush]

So what time does it usually fix itself?

All three days it failed at 10:05 UTC +- few minutes. Then recover for few minutes around 10:30, then fall down again and looks like finally started working after 11:00. Of course tried to restart everything and watch everything, including possibly malicious traffic on external interfaces (getwork and bitcoin p2p), but everything looks normally. It is just incredibly laggy at this time, so 99% of requests fail on timeouts.

It might be related to weekend infrastructure update, when all server plans were updated (and severs were physically moved to another machines). I requested moving all servers to another locations, maybe it helps. If it will repeat tomorrow, I'll change provider...

[slush]

yeah thats weird bro. I have been watching it too every day around this time.
Are you sure the server does not do any maintenance tasks at this time of day?

There are of course no cronjobs on my servers. But as servers are virtual, it might be caused by heavy maintenance of *other* machines.

Or perhaps your uplink does?

Uplink works well, I see normal traffic here.

[slush]

It may be helpful that while the server malfunctions refreshing the web page results often: "Error 502 - bad gateway"

Edit: or "Error 504 - gateway timeout."

I know that, it was because I tried to restart everything, including website. Unfortunately nothing helped :-/

[slush]

Reward for block 4753 isn't correct because block was finished during outage, I will recalculate it later...

[MiningBuddy]

yeah thats weird bro. I have been watching it too every day around this time.
Are you sure the server does not do any maintenance tasks at this time of day?

There are of course no cronjobs on my servers. But as servers are virtual, it might be caused by heavy maintenance of *other* machines.

Or perhaps your uplink does?

Uplink works well, I see normal traffic here.

Slush, forgive me for speaking out of line here but isn't it about time you invested in some real dedicated machines for your pool and not cheap VPS solutions?

[slush]

Slush, forgive me for speaking out of line here but isn't it about time you invested in some real dedicated machines for your pool and not cheap VPS solutions?

This is of course valid question. I'm Linode customer for many years and I have so far great experience with them. I don't think it is 'cheap VPS solution', it is just 'VPS solution'. Problems happen everywhere, for example one my real rack server catched up by fire before few months. 'real == better' isn't valid in every situation.

VPS is very good solution for application like pool, because I can scale it as I wish. Pool went from single server with 512MB RAM (on 16. December) to many server infrastructure with standalone database, balancer with possible IP failover, many backends where I can disconnect any of them without pool outage and so on. Everything that with 15 minute planned outage because of database migration. In case of rising demand, I can make pool infrastructure 10x stronger by few clicks. This is not possible with real servers.

I hope that we'll find reason for those issues, and I'm really sorry for it; we're trying to do the best. But if we won't find any trouble, I'm of course ready to migrate to another provider to give best possible pool experience.

[cablepair]

your doing a great job slush

right now your pool has the best size/speed ratio out there

deepbit is super fast but they are too big
you cant get a decent reward per round because of all the competition for the rewards and the fees make it even worse

the smaller guys have much bigger rewards per round but they take too long to solve a block

right now your size and speed are perfect, I am actually going to be writing an article with comparison of all this , I just hope your pool does not get too big

deepbit is way too big, not good for the network or for the miners

[oVPN]

hey Slush, if you need some additional power, i could help you with. have a nice 24g ram server with i7-960 cpu running in ukraine for my own testing purposes, so its most of the time fully idle. its virtualized with openvz and i could give you some mirror with your own IP, cores and space what you need with 2-4G ram, if you want: send me pm or for contact details, icq or jabber prefered

[allinvain]

Slush, forgive me for speaking out of line here but isn't it about time you invested in some real dedicated machines for your pool and not cheap VPS solutions?

This is of course valid question. I'm Linode customer for many years and I have so far great experience with them. I don't think it is 'cheap VPS solution', it is just 'VPS solution'. Problems happen everywhere, for example one my real rack server catched up by fire before few months. 'real == better' isn't valid in every situation.

VPS is very good solution for application like pool, because I can scale it as I wish. Pool went from single server with 512MB RAM (on 16. December) to many server infrastructure with standalone database, balancer with possible IP failover, many backends where I can disconnect any of them without pool outage and so on. Everything that with 15 minute planned outage because of database migration. In case of rising demand, I can make pool infrastructure 10x stronger by few clicks. This is not possible with real servers.

I hope that we'll find reason for those issues, and I'm really sorry for it; we're trying to do the best. But if we won't find any trouble, I'm of course ready to migrate to another provider to give best possible pool experience.

Slush have you considered switching to a provider such as Rackspace or Softlayer? I do believer they have a cloud hosting solution which may be perfect for the pool - you can scale the amount of server hardware you need.

Another provider to consider may be iWeb. They're Canadian based, but when I had a dedicated server with them I found them to have a very reliable network. As far as I know they have cloud type of solution now too.

Maybe you can even convince some of these provider to accept payment in bitcoins! That would be super cool.

[slush]

Slush have you considered switching to a provider such as Rackspace or Softlayer?

Yes, I'm thinking about them as alternative to Linode.

Another provider to consider may be iWeb. They're Canadian based, but when I had a dedicated server with them I found them to have a very reliable network.

We have ~30 servers in iWeb and as far as I can say, their support is horrible. For example - waiting two days for connecting console to server? Really?

[woah]

If you ever decide to change hosting, I would be very happy if it stays in Europe, ( I take it they are in London now ? ) your pool is the only one with which I have acceptable latency. All the other ones are far from here ( France ) , and I get a lot of stale shares with them.

[urtur]

Wow, what a fast block:
#           Block found                   Duration   Total shares    Your reward    Block #   
4761    2011-05-24 14:12:58    0:00:27         2086              xxxxxx           126326