08-14-2003 4:15pm Onwards, Continent wide power issues
Moderator: Admins
-
- Site Admin
- Posts: 711
- Joined: Wed Jun 12, 2002 5:57 pm
- Location: Toronto, Ontario
- Contact:
Yeah,ZeroFill wrote:yes, they all take ~10sec
I'm seeing an almost instant resolve of your DNS here, but thats because we've been checking on and off for the last day, it's definatly cached by now. I've run out of new servers to check through, but basically once your ISP stops taking 12 seconds to resolve your RDNS from external queries, your ftp will stop taking 12+ seconds to connect.
This shouldn't affect any other performance (eg files will transfer fast still, etc.), and is most likely related to your power (I know Cogeco's servers have been flooded since this power issue, our own RDNS for my home cable has been noticably slower as they're performing a load of system maintenance, etc.).
i doubt we have a power issue as i am on the west coast -- we werent affected by the power.
what can i tell my isp to fix this problem? i'm not sure if they will even look into it because all other ftps work instantly. i still dont know why other ftp servers connect instantly but prioritycolo's takes 12 secs. (it's hard to believe it is my isp because other ftp servers work fine)
also, like you said, my ip is cached, so why does it take a long time to log in?
what can i tell my isp to fix this problem? i'm not sure if they will even look into it because all other ftps work instantly. i still dont know why other ftp servers connect instantly but prioritycolo's takes 12 secs. (it's hard to believe it is my isp because other ftp servers work fine)
also, like you said, my ip is cached, so why does it take a long time to log in?
-
- Site Admin
- Posts: 711
- Joined: Wed Jun 12, 2002 5:57 pm
- Location: Toronto, Ontario
- Contact:
Ahh ok, my bad for making assumptions.ZeroFill wrote:i doubt we have a power issue as i am on the west coast -- we werent affected by the power.
what can i tell my isp to fix this problem? i'm not sure if they will even look into it because all other ftps work instantly. i still dont know why other ftp servers connect instantly but prioritycolo's takes 12 secs. (it's hard to believe it is my isp because other ftp servers work fine)
also, like you said, my ip is cached, so why does it take a long time to log in?
Simply state to your ISP that external servers are taking over 5 seconds to do RDNS lookups on your IP address, and you want to know why. Theres no 100% certainty I'm right on this one mind you, but I don't see anything else causing this slowdown, and it's the co-relation you have with the snickers.org box (which also took over 10 seconds to do a RDNS lookup).
I'm uncertain if ProFTPD takes IP cache into consideration when it does lookups, I know little about the internal workings of the program other then what it logs, thus it's arguable that it may use IP cache, or may not, I honestly couldn't tell you.
i tried doing the time host thing again on another shell i have and these are my results.
18:49:46 lawicdco:~
$ time host 24.205.164.195
195.164.205.24.IN-ADDR.ARPA domain name pointer 24-205-164-195.wc-eres.charterpipeline.net
real 0m0.138s
user 0m0.000s
sys 0m0.003s
18:50:03 lawicdco:~
$ time host 24.205.164.195
195.164.205.24.IN-ADDR.ARPA domain name pointer 24-205-164-195.wc-eres.charterpipeline.net
real 0m0.002s
user 0m0.000s
sys 0m0.002s
why would it take 10sec off your box, yet .138sec off another box? the other shell i tested this has no relation to my isp. how could i convince my isp into looking into the problem if i cannot reproduce the lag off a 3rd party box? i'm starting to think there's a link/router problem from the US to your location.
[edit]
ok. now this is really confusing. that 3rd party box connects to you fine
18:57:22 lawicdco:~
$ time ftp www5.pcdc.net
220 ProFTPD 1.2.8 Server (ProFTPD) [66.11.162.63]
Name (www5.pcdc.net:lawicdco): ^C
real 0m1.161s
user 0m0.000s
sys 0m0.003s
and yet i cannot reproduce the 10sec lag to my home box. arg!!!
18:49:46 lawicdco:~
$ time host 24.205.164.195
195.164.205.24.IN-ADDR.ARPA domain name pointer 24-205-164-195.wc-eres.charterpipeline.net
real 0m0.138s
user 0m0.000s
sys 0m0.003s
18:50:03 lawicdco:~
$ time host 24.205.164.195
195.164.205.24.IN-ADDR.ARPA domain name pointer 24-205-164-195.wc-eres.charterpipeline.net
real 0m0.002s
user 0m0.000s
sys 0m0.002s
why would it take 10sec off your box, yet .138sec off another box? the other shell i tested this has no relation to my isp. how could i convince my isp into looking into the problem if i cannot reproduce the lag off a 3rd party box? i'm starting to think there's a link/router problem from the US to your location.
[edit]
ok. now this is really confusing. that 3rd party box connects to you fine
18:57:22 lawicdco:~
$ time ftp www5.pcdc.net
220 ProFTPD 1.2.8 Server (ProFTPD) [66.11.162.63]
Name (www5.pcdc.net:lawicdco): ^C
real 0m1.161s
user 0m0.000s
sys 0m0.003s
and yet i cannot reproduce the 10sec lag to my home box. arg!!!
-
- newbie
- Posts: 1
- Joined: Tue Aug 19, 2003 10:04 pm
-
- Site Admin
- Posts: 711
- Joined: Wed Jun 12, 2002 5:57 pm
- Location: Toronto, Ontario
- Contact:
Thats exactly what makes troubleshooting a problem like this so difficult, and why you may get varrying results. We test from 3-4 servers, one internal, the rest external, and base our conclusions on that, often they're right, sometimes they're totally wrong. Best case suggestion, wait and see if it disappears, I don't believe we changed anything before, I think it just went away on it's own did it not?
yeah, the login-lag did seem to just dissapear but this was fixed after you posted the complete switch over to istop or peer1. are the routers still configured the same way it was before the blackout?
i thought it was the switchover that fixed the login-lag and that new configuration w/ the routers et-all mitigated the problem.
i thought it was the switchover that fixed the login-lag and that new configuration w/ the routers et-all mitigated the problem.
-
- Site Admin
- Posts: 711
- Joined: Wed Jun 12, 2002 5:57 pm
- Location: Toronto, Ontario
- Contact:
Not too likely. The routers are being tweaked 24x7, but your login process is no different from download. If you had a poor route to us, you'd see it with everything you do, uploads/downloads would go dead slow, email would take 20+ seconds to retrieve, traceroutes would be horrid, etc. It's not a routing issue, as you'd be complaining about a lot more before your FTP login took 10 seconds to login.ZeroFill wrote:yeah, the login-lag did seem to just dissapear but this was fixed after you posted the complete switch over to istop or peer1. are the routers still configured the same way it was before the blackout?
i thought it was the switchover that fixed the login-lag.