Thread: [OpenSTA-users] IO Failed for VU 1-1 on connection 0/6 with error 64:(The specified network name
Brought to you by:
dansut
Menu
▾
▴
opensta-users
|
From: Rajendran C. <raj...@ho...> - 2007-03-12 10:18:00
|
Hello All, I have been doing performance testing of a website, and when i try 100 VU's, i get error " IO Failed for VU 1-1 on connection 0/6 with error 64: ( The specified network name is no longer available, )., i have enough memory and network resource on both the client and server side. i researched a lot on this error, also saw a posting on the same issue, but no answer, however i know this is a system error 64, which could occur if any issues on the sockets or if the nic card is not able to send and recieve traffic. does this error arise when there is more load on the webserver ? if so what is the response code it should return, Pls help (also when i tried to download the whole website and put in our lan network and checked putting load, for as less as 25 users, i started getting error 64 from opensta, so while i was loading from opensta, i opened a browser and tried to connect to the localhost site, and it gave client error "403.9" response code, too many users accessing, but when i test the real production server (webserver has a Proxy server enabled), it does not give response code '403') Thanks Rajendran c _________________________________________________________________ Spice up your IM conversations. New, colorful and animated emoticons. Get chatting! http://server1.msn.co.in/SP05/emoticons/ |
|
Re: [OpenSTA-users] IO Failed for VU 1-1 on connection 0/6 with
error 64:(The specified network name
From: Dan D. <ddo...@me...> - 2007-03-12 14:16:34
|
Rajendran Wrote: >I have been doing performance testing of a website, and when i try 100 VU's, >i get error " IO Failed for VU 1-1 on connection 0/6 with error 64: ( The >specified network name is no longer available, )., Rajendran, My experience and understanding of this error is that it is "load related", and the moral equivalent of an http error 500 -- even though an error 500 may not actually be received. Namely, the web server is overwhelmed and is rejecting connections. We see these a lot at a certain load ramp point when the target system starts to get overloaded. But, I would love to be enlightened further -- especially if I am mistaken -- by DanS or Olaf or some other opensta guru... ..Dan www.mentora.com |
|
Re: [OpenSTA-users] IO Failed for VU 1-1 on connection 0/6 with
error 64:(The specified network name
From: Olaf K. <ok...@ab...> - 2007-03-12 14:32:00
|
Rajendran Chandrasekaran schrieb: > Hello All, > > I have been doing performance testing of a website, and when i try 100 VU's, > i get error " IO Failed for VU 1-1 on connection 0/6 with error 64: ( The > specified network name is no longer available, )., [...] > does this error arise when there is more load on the webserver ? > if so what is the response code it should return, If there is no connection to the webserver, there will be no HTTP protocol exchange on the line, thus no response code. And yes, Error 64 is the Winsock error, promoted directly from the underlying network api, more documentation about this error code may be found at Microsofts Windows API documentation (or at google). > (also when i tried to download the whole website and put in our lan network > and checked putting load, for as less as 25 users, i started getting error > 64 from opensta, so while i was loading from opensta, i opened a browser and > tried to connect to the localhost site, and it gave client error "403.9" > response code, too many users accessing, but when i test the real production > server (webserver has a Proxy server enabled), it does not give response > code '403') I'm not aware of any non-integer response codes - e.g. 403.9 to me is unknown. I believe "too many users accessing" is a pretty obvious error message. I suppose, there are too many users that currently access your server. If it doesn't happen when the server is configured differently, maybe the configuration makes a difference? Sorry, to me it's not clear, what you are asking here... Cheers, Olaf |
|
From: Bernie V. <Ber...@iP...> - 2007-03-12 17:28:48
|
> Hi All, > Can you please tell me how to increase the timeout of OpenSTA, I > understand it is currently 1 minute. > Chris, I think everyone has been very patient and helpful with your requests here. We want you to be successful using OpenSTA. There comes a point however when it becomes apparent that you are not trying very hard to find answers for yourself. I personally gave you advise on where to read about this on 3/9. Try goggling 'opensta timeout' and read the first response. It will point you to: http://portal.opensta.org/modules.php?op=modload&name=phpWiki&file=index&pagename=PlaybackRequestTimeout Besides using search engines, please take the time to read the FAQ at http://portal.opensta.org/modules.php?op=modload&name=phpWiki&pagename=FrequentlyAskedQuestions&file=index and the online doc at http://opensta.org/docs/ If you don't have the time to research this yourself then you should consider hiring someone to help accelerate your project. All the best, -Bernie (www.iPerformax.com) |
|
From: Bernie V. <Ber...@iP...> - 2007-03-12 14:35:55
|
> I have been doing performance testing of a website, and when i try 100 > VU's, > i get error " IO Failed for VU 1-1 on connection 0/6 with error 64: ( The > specified network name is no longer available, )., > Rajendran, I have encountered this error under two different situations. The first was when the system under test had two NICs. In that case, disabling one of the NICs cured the problem. The other time it was related to a network bottleneck when running multiple OpenSTA boxes trying to emulate upwards of 3,000 virtual users. Each OpenSTA box (and the NIC on it) had adequate capacity but the sum of the network traffic saturated one of the routers. You really have to dig into the network configuration to be sure there is adequate capacity. You might try pinging the system under test from the OpenSTA server when the error 64 occurs to see how the network is responding or if you get a similar error outside of OpenSTA. Please post a reply if you solve this problem as it has been reported here a number of times and there is no clear cause or resolution that I am aware of. -Bernie (www.iPerformax.com) |
|
From: Daniel S. <da...@Op...> - 2007-03-12 18:01:23
|
System Error 64: "The specified network name is no longer available" Is as the message suggests a "system" error and OpenSTA is just passing the info - it is unfortunately one of those inconclusive errors that really just tells you something was exhausted or broke - in this specific case the 'something' is to do with the network. Bernie Velivis wrote: > I have encountered this error under two different situations. The > first was when the system under test had two NICs. In that case, > disabling one of the NICs cured the problem. Although there is no bug associated with it, it is known that there are 'issues' running the load generator on machines with multiple NICs - I didn't remember that this presented itself as an 'error 64' though. Seems a strange way to present itself, but maybe not when you read the theory below ... > The other time it was related to a network bottleneck [snip] I think you need say no more here :) I suggest that as we can assume that load testing is being done then something is being exhausted by your load generator - this might be a system resource or a network infrastructure resource; I'm afraid that only the person experiencing this is going to be able to find out which. Check your system error logs and maybe try to track some of the potentially exaustable resources using PerfMon - if you have a managed switch or router then you may be able to track stuff there as well. So, given this, here's my theory on the multiple NIC problem: maybe having multiple NICs isn't really a problem, maybe it is just that having more than 1 NIC allows what is ever being exhausted to occur quicker/easier - just having a single NIC acts as a kindof throttle to stop things becoming overwhelmed and breaking? Just an idea. Related: did recording on the machine with multiple NICs work fine before you disabled one of them. Cheers /dan -- Daniel Sutcliffe <Da...@Op...> OpenSTA part-time caretaker - http://OpenSTA.org/ |
|
From: Bernie V. <Ber...@iP...> - 2007-03-12 18:49:28
|
>> The other time it was related to a network bottleneck > [snip] > I think you need say no more here :) Hi Dan, The network bottleneck I alluded to occured when trying to emulate 3000 users over 6 (or so, its been a while) OpenSTA load generators. The load generation machines were configured with a single gigabit ethernet controllers but the controllers, set to autodetect speed, were only running at 100 100Mb. Each OpenSTA server was averaging about 20 Mbit/sec over it's single ethernet controller, implying each was running well below its theoretical limit. CPU, memory, and disk activity on the OpenSTA servers were low enough to not be a factor. The network topology had a gigabit backbone, but after digging under the covers and asking the same question 10 times (I was very curious why they had fallen back to 100 Mbit mode), I finally learned that "OH! Thats right, all those OpenSTA boxes eventurally go through a single 100 Mbit switch to get to their SUTs". So while each of the servers (load generators and servers under test) were each runing at well below 100 Mbit/sec, the sum of the network traffic had exceeded the capacity of the 100 Mbit switch burried somewher in the route between the OpenSTA servers and the SUTs. When we were receiving "The specified network name is no longer available" errors, it was during a period where one 100 Mbit network switch in the route to the SUT was trying to service in excess off 100 Mbit/sec woth of traffic sustained over a long period. -Bernie |
|
From: Osoata, C. <Chr...@at...> - 2007-03-12 15:34:48
|
Hi All, Can you please tell me how to increase the timeout of OpenSTA, I understand it is currently 1 minute. Thanks Chris This email and any attached files are confidential and copyright protected.= If you are not the addressee, any dissemination of this communication is s= trictly prohibited. Unless otherwise expressly agreed in writing, nothing s= tated in this communication shall be legally binding. The ultimate parent company of the Atkins Group is WS Atkins plc. Register= ed in England No. 1885586. Registered Office Woodcote Grove, Ashley Road, = Epsom, Surrey KT18 5BW. Consider the environment. Please don't print this e-mail unless you really = need to.=20 |
|
From: Daniel S. <da...@Op...> - 2007-03-12 17:31:16
|
Christabel Osoata wrote: > Can you please tell me how to increase the timeout of OpenSTA, I > understand it is currently 1 minute. By reading the FAQ ;) Olaf even gave you the link to the correct page in the FAQ in the first reponse to your question that led to this question ... Cheers /dan -- Daniel Sutcliffe <Da...@Op...> OpenSTA part-time caretaker - http://OpenSTA.org/ |
|
Re: [OpenSTA-users] IO Failed for VU 1-1 on connection 0/6 with
error 64:(The specified network name
From: Daniel S. <da...@Op...> - 2007-03-12 18:01:14
|
Rajendran Chandrasekaran wrote: > (also when i tried to download the whole website and put in our lan > network and checked putting load, for as less as 25 users, i started > getting error 64 from opensta, so while i was loading from opensta, i > opened a browser and tried to connect to the localhost site, and it > gave client error "403.9" response code, too many users accessing, > but when i test the real production server (webserver has a Proxy > server enabled), it does not give response code '403') My guess as to the cause of this, is that your LAN installed Web server is IIS on a "Workstation" version of Windows (not "Server"). These IIS installs are limited on the number of connections they allow. Check your System error log for messages. Cheers /dan -- Daniel Sutcliffe <Da...@Op...> OpenSTA part-time caretaker - http://OpenSTA.org/ |
Thanks for helping keep SourceForge clean.
X