ny2cafuse (OP)
Legendary
Offline
Activity: 1582
Merit: 1002
HODL for life.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 03:38:23 PM Last edit: December 13, 2013, 04:58:24 PM by ny2cafuse |
|
Well here we go again ![Angry](https://bitcointalk.org/Smileys/default/angry.gif) . Ok, so this definitely has to be a traffic issue at this point. The CPU and memory graphs, as well as running TOP, on the frontend and DB servers all come back with little to no usage. I'm going to be putting in a service inquiry to the server provider to find out why they can't seem to handle the load. In the meantime I've asked the whale from China to leave. I know he was pulling blocks for us at an awesome speed, but I'll take a steady pool over the lag we're getting any day. I sent him an email this morning, so we'll see what happens today. I'm sorry guys ![Sad](https://bitcointalk.org/Smileys/default/sad.gif) . I was banking on this upgrade fixing things. In the meantime, I guess I'm going to start having to look for another hosting provider with a faster connection. Fudge-knuckle.
|
Community > Devs
|
|
|
vesperwillow
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 03:42:11 PM |
|
I wish I had access to cheap processing like in China.. lol.
What provider are you using? It's definitely seems like a pipe issue.
Also, is this likely why the dashboard doesn't show all of my miners? They say they're connected and providing shares, but the dashboard isn't crediting me. At times it's showing no miners.
|
|
|
|
ny2cafuse (OP)
Legendary
Offline
Activity: 1582
Merit: 1002
HODL for life.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 04:03:20 PM Last edit: December 14, 2013, 12:15:45 AM by ny2cafuse |
|
I wish I had access to cheap processing like in China.. lol.
What provider are you using? It's definitely seems like a pipe issue.
Also, is this likely why the dashboard doesn't show all of my miners? They say they're connected and providing shares, but the dashboard isn't crediting me. At times it's showing no miners.
I'm using Digital Ocean. Support ticket went in just now, so I'll see what they say. I think the worker issue is definitely related to this slowdown. Right now, that whale is throwing exactly 119 workers at the pool. That's wrapping up the workers table with his workers, and I think that's where the issue with the missing worker stats lies. Make sure your workers are showing up with hashrates in the workers page. I read somewhere where the pool ops were turning this off on the dashboard to save resources. I might try that for a day or two later to see what's going on. If the whale isn't gone by the end of the day, I'm going to email him again using google translate. -Fuse
|
Community > Devs
|
|
|
AFK.MPEG
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 72
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 05:57:32 PM |
|
How big is your server upload? And do you have loadbalancing software?
|
|
|
|
ny2cafuse (OP)
Legendary
Offline
Activity: 1582
Merit: 1002
HODL for life.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 06:51:16 PM |
|
How big is your server upload? And do you have loadbalancing software?
Server upload? You mean bandwidth on the upload connection or the size of the server? The back-end mysql cluster is load-balanced via mysqlnd_ms, which is supported on the MPOS installation. I can see traffic between the cluster, so I know it is working properly. I need to trace the bottleneck and find out if there is a fix. I'll be working on this throughout the weekend. I logged into the Hashfaster pool and saw that they're in the same boat as us when they hit ~150MH. I might drop a line to a few helpful pool ops and see if they have any suggestions. If anyone is a server pro, PM me if you have any advice. -Fuse
|
Community > Devs
|
|
|
vesperwillow
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 06:54:46 PM |
|
How big is your server upload? And do you have loadbalancing software?
Server upload? You mean bandwidth on the upload connection or the size of the server? The back-end mysql cluster is load-balanced via mysqlnd_ms, which is supported on the MPOS installation. I can see traffic between the cluster, so I know it is working properly. I need to trace the bottleneck and find out if there is a fix. I'll be working on this throughout the weekend. I logged into the Hashfaster pool and saw that they're in the same boat as us when they hit ~150MH. I might drop a line to a few helpful pool ops and see if they have any suggestions. If anyone is a server pro, PM me if you have any advice. -Fuse If another pool is having the same issues, that's a starting point. Are they on the same service provider? If so there's a possibility, but if not it could point to an issue with the backend of the coin itself. I mined some cubits when they first hit, and it was fine with low power but when I threw a few hundred at it, the server choked. Others reported similar issues. Pool operator kept saying nothing was wrong, but it ended up being a bug in the coin/code.
|
|
|
|
24Kilo
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 06:55:00 PM |
|
Fuse,
Since the whale left, query times are sub-100ms, averaging around 60ms.
If he was running 119 workers, then he must be using GPU's and running a instance for GPU because he is averaging 700Kh/s for each worker. It is a shame to see him leave, so maybe suggest that he combine GPU's so he needs less workers or maybe the he ramp up his miners over the course of a few days to provide you a chance to work out what is needed to handle the load.
HashFaster's pool is running around 130Mh/s, but according to the pool register the largest miner is about 10Mh/s and the load is spread more evenly across the top 15.
Not sure what I can do to help, but let me know what you need and I may be able to contribute something.
24Kilo
Edit - query times for HashFaster's pool as I type this are 3000ms plus... fluctuating between 2500 to over 5000 at times.
|
|
|
|
24Kilo
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 07:12:01 PM |
|
...when I threw a few hundred at it, the server choked... You sound like a bit of a whale yourself? A few hundred?! The pool seems best both in yield and query times around 60Mh/s...
|
|
|
|
vesperwillow
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 07:14:28 PM |
|
...when I threw a few hundred at it, the server choked... You sound like a bit of a whale yourself? A few hundred?! The pool seems best both in yield and query times around 60Mh/s... Noo, not quite there /yet/ lol. I'm trying though. I was referring to an SHA256 coin, so it was GH.
|
|
|
|
ny2cafuse (OP)
Legendary
Offline
Activity: 1582
Merit: 1002
HODL for life.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 07:15:38 PM |
|
Fuse,
Since the whale left, query times are sub-100ms, averaging around 60ms.
If he was running 119 workers, then he must be using GPU's and running a instance for GPU because he is averaging 700Kh/s for each worker. It is a shame to see him leave, so maybe suggest that he combine GPU's so he needs less workers or maybe the he ramp up his miners over the course of a few days to provide you a chance to work out what is needed to handle the load.
HashFaster's pool is running around 130Mh/s, but according to the pool register the largest miner is about 10Mh/s and the load is spread more evenly across the top 15.
Not sure what I can do to help, but let me know what you need and I may be able to contribute something.
24Kilo
Edit - query times for HashFaster's pool as I type this are 3000ms plus... fluctuating between 2500 to over 5000 at times.
I'm currently consulting with a sysadmin on how to optimize the back-end settings. I'm hoping that helps, because I can't see having to migrate again. I felt bad asking the whale to leave, but I'd rather it be stable right now for you guys while I figure this out. Besides, I'm hoping I can get this working correctly so I can pull Mark into the pool with his incoming 75MH farm he's building. If I can't get this figured out, I'm going to cry ![Cry](https://bitcointalk.org/Smileys/default/cry.gif)
|
Community > Devs
|
|
|
vesperwillow
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 07:24:37 PM |
|
It sounds like we need to get together and...
..wait for it
Tag-Team it!
If we can help let us know.
|
|
|
|
AFK.MPEG
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 72
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 08:35:02 PM |
|
How big is your server upload? And do you have loadbalancing software?
Server upload? You mean bandwidth on the upload connection or the size of the server? The back-end mysql cluster is load-balanced via mysqlnd_ms, which is supported on the MPOS installation. I can see traffic between the cluster, so I know it is working properly. I need to trace the bottleneck and find out if there is a fix. I'll be working on this throughout the weekend. I logged into the Hashfaster pool and saw that they're in the same boat as us when they hit ~150MH. I might drop a line to a few helpful pool ops and see if they have any suggestions. If anyone is a server pro, PM me if you have any advice. -Fuse Yeah same thing. Upload, bandwifth same thing. how many connections does your server have? 1Gbit? 2 Gbit? and how big is your hosts upload speed? 100mbit or Gbit? Your software is it running on a Harddisk? Because when i host heavy SQL applications on our LAN-party everything is loaded in a RAMdisk to handle 200 sql connections. Otherwise it bottleneck like hell. Maybe a idea?
|
|
|
|
ny2cafuse (OP)
Legendary
Offline
Activity: 1582
Merit: 1002
HODL for life.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 08:46:45 PM |
|
Yeah same thing. Upload, bandwifth same thing. how many connections does your server have? 1Gbit? 2 Gbit? and how big is your hosts upload speed? 100mbit or Gbit?
Your software is it running on a Harddisk? Because when i host heavy SQL applications on our LAN-party everything is loaded in a RAMdisk to handle 200 sql connections. Otherwise it bottleneck like hell. Maybe a idea?
PM incoming.
|
Community > Devs
|
|
|
AFK.MPEG
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 72
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 08:59:52 PM |
|
Yeah same thing. Upload, bandwifth same thing. how many connections does your server have? 1Gbit? 2 Gbit? and how big is your hosts upload speed? 100mbit or Gbit?
Your software is it running on a Harddisk? Because when i host heavy SQL applications on our LAN-party everything is loaded in a RAMdisk to handle 200 sql connections. Otherwise it bottleneck like hell. Maybe a idea?
PM incoming. Omg SPAM ![Shocked](https://bitcointalk.org/Smileys/default/shocked.gif)
|
|
|
|
ny2cafuse (OP)
Legendary
Offline
Activity: 1582
Merit: 1002
HODL for life.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 09:24:41 PM |
|
Upon further investigation, I have found a lot of references to the querytime issues on the MPOS github. TheSerapher's recommendation is disabling the live dashboard. Apparently this causes an insane amount of load on the DB. There is talk about changing the way these statistics are cached, so I think for now the best bet is to shut it off. I know it's snazzy and all, but if it improves server response for now, I'm going to do it.
So bring on the whales and let's see what happens.
|
Community > Devs
|
|
|
vesperwillow
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 09:35:36 PM |
|
Upon further investigation, I have found a lot of references to the querytime issues on the MPOS github. TheSerapher's recommendation is disabling the live dashboard. Apparently this causes an insane amount of load on the DB. There is talk about changing the way these statistics are cached, so I think for now the best bet is to shut it off. I know it's snazzy and all, but if it improves server response for now, I'm going to do it.
So bring on the whales and let's see what happens.
Snazzy is fine, but faster server is better ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) I switched back to you, let's hope for the best.
|
|
|
|
vesperwillow
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 10:21:02 PM |
|
Not to throw gas on the fire, but I think in the rebuild the auto-logout reverted back to the shorter time.
|
|
|
|
ny2cafuse (OP)
Legendary
Offline
Activity: 1582
Merit: 1002
HODL for life.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 10:39:20 PM |
|
Not to throw gas on the fire, but I think in the rebuild the auto-logout reverted back to the shorter time.
![](https://ip.bitcointalk.org/?u=http%3A%2F%2Fct.fra.bz%2Fol%2Ffz%2Fsw%2Fi57%2F5%2F10%2F8%2Ffrabz-like-bro-throwing-gas-in-a-fire-is-a-good-idea-anytime-157cf3.jpg&t=665&c=XrRQu5XI-wrzZQ) In all seriousness, I did a couple of Apache restarts which probably reset your cookie. Let me know if it continues to happen.
|
Community > Devs
|
|
|
24Kilo
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 11:12:22 PM Last edit: December 14, 2013, 12:03:07 AM by 24Kilo |
|
Fuse,
I was very happy with the old web interface that did not have all the bells and whistles. A fast and responsive server is first priority. And we need to get the pool back to 100% efficiency again.
Again many thanks for all your hard work.
Just thought I would ask, does anyone here know where to get PCI-E riser cables in Australia other than eBay? Preferably in Sydney.
Thanks!
24Kilo
|
|
|
|
ny2cafuse (OP)
Legendary
Offline
Activity: 1582
Merit: 1002
HODL for life.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
December 13, 2013, 11:39:53 PM |
|
Fuse,
I was very happy with the old web interface that did not have all the bells and whistles. A fast and responsive server is first priority. And we need to get the pool back to 100% efficiency again.
Again many thanks for all your hard work.
Just thought I would ask, does anyone here know where to PCI-E riser cables in Australia other than eBay? Preferably in Sydney.
Thanks!
24Kilo
For now I'm going to stick with the MPOS layout with the live stats disabled. I can get why they were bombing the server now. They basically calculated everyone's stats every 5-10 seconds, so the whale came on and the stats shit the bed. So unless I migrate over to a dedicated(aka pricey as shit) server solution, which I probably won't do at this time, I'll just leave the stats disabled. I'd rather take the money the pool makes and give it back to the miners in jackpots. People can hit F5 or click on the dashboard link over and over if they have to have live stats. I haven't tested to see if that breaks API calls, so anyone using them, please let me know. If they still work, they can use that too. As far as risers go, does Amazon have an AU hub for things other than books? Other option is to buy one via marketplace here. There are a lot of Aussie miners hanging around in there it seems. -Fuse
|
Community > Devs
|
|
|
|