- WLUG - wlug.mailman3.com

Good price for Nokia N810

by Clint Moyer

With word out that the N900 will be released RSN, the price of the N810 has dropped. $180 at buy.com. http://www.buy.com/prod/nokia-n810-internet-tablet-full-pull-out-qwerty-key…

June 15, 2009

Re: [Wlug] Time and Date Stamp on Traceroute?

by E Johnson

Thanks guys > Traceroute is not the proper tool to ascertain the bandwidth available. Chuck, I am not using traceroute to determine bandwidth Just using that at selected times because it shows all the hops to the designted IP in milliseconds, which often reveals a big delay in one of the hops. Have to show the ISP that the bottleneck is in their stations and not farther out in the internet. > Even traceroute packet loss doesn't necessarily indicate a problem, since the lower priority of the ICMP or UDP traceroute packets means they might get dropped intentionally if the router is busy processing real traffic. Where is this lower priority determined? > If the ISP is asking you for a traceroute... ISP is not asking me for a traceroute. ISP is asking me to prove how poorly their 20 mb/s internet (which I am testing) "works." They want me to buy this service permanently at a $25 increase over my usual rate for the 10 Mb/s service I usually subscribe to --which isn't usually anywhere near 10 Mb/s either. Meanwhile, I want them to fix the poor-bandwidth problem we've had here for 4 1/2 years. Also meanwhile, during the times of worst bandwidth here, I find very-long latency times in the ISP's closest hops here. For whatever reason these are happening, I think they should see that. What I am doing is trying to assemble as many tools with as much info as possible, in order to communicate this problem to them. > If you really want to test the performance of your connection, you need two well-defined endpoints, such as your PC and a server at your ISP, both running software designed for this purpose. I am also using the server at the ISP which they recommended, using a little gui they have for this purpose. Typical bandwidth today between 12 PM and 2 PM was running between 9 Mb/s and 14 Mb/s, on the service that's supposed to give me 20 Mb/s. Thanks for the other recommendations & ideas, I will check them out & read the recommended pages. Liz J

June 6, 2009

Re: [Wlug] Time and Date Stamp on Traceroute?

by Chuck Anderson

On Sat, Jun 06, 2009 at 01:42:38PM -0400, E Johnson wrote: > Does anyone know a quick command I can use to add a time and date stamp to > traceroute? Is this not enough? date; traceroute foo > I need a quick app (nothing that generates a significant amount of network > traffic) --where I can enter one command in terminal which will display a > printout of a traceroute, to show the ISP or to show the tech in real-time > when he comes out. The one I send the ISP needs the time/date in order to be > effectively convincing to them. If the ISP is asking you for a traceroute, then they would know that no standard traceroute programs out there actually have a built-in date/time stamp functionality. "tracert" and "pathping" on Windows don't add a time stamp that I know of. Running "date" followed by "traceroute" should be enough. Put it in a shell script so you can invoke it as a single command if you want. There is a nicer traceroute for Linux called "Matt's Traceroute" or mtr. It prints some additional information like packet loss as well as a date/time at the top of the screen. It uses an ncurses full-screen interactive interface. > We're supposed to get 20,000 kb/s (up) here, and quite often, we're > not, even with very low internet use in the PCs I'm running here. > Even the download speed's choking... seems like something big right > close is often are clogging the pipe. Traceroute I'm doing is only > going from here to Cambridge. Traceroute is not the proper tool to ascertain the bandwidth available. Traceroute tells you which route the packets take, and the "latency" given by each hop is not an accurate indicator of anything really. Traceroute packets are often processed in software by the routers along the path, often at a lower priority and rate limited compared to "real" traffic that they route, so the numbers they give for latency are not useful, and certainly can't be used to tell you the bandwidth of a connection. Even traceroute packet loss doesn't necessarily indicate a problem, since the lower priority of the ICMP or UDP traceroute packets means they might get dropped intentionally if the router is busy processing real traffic. If you really want to test the performance of your connection, you need two well-defined endpoints, such as your PC and a server at your ISP, both running software designed for this purpose. Common software used for performance testing is ttcp, netperf, iperf. Here is some more useful information about the topic: http://en.wikipedia.org/wiki/Bandwidth_test

June 6, 2009

Re: [Wlug] Time and Date Stamp on Traceroute?

by J.R. Mauro

On Sat, Jun 6, 2009 at 1:42 PM, E Johnson<iris.gates(a)gmail.com> wrote: > Does anyone know a quick command I can use to add a time and date stamp to > traceroute? Traceroute-nanog. It may be hard to find the original source. Try looking for the debian package online, and on the right hand side of the website, there should be a source tarball. I tried finding the "aces.com" site where it's supposed to be available, but no such luck. I imaging the debian way is the way to go. > Man page for traceroute doesn't seem to have that option. > > Or is there a similar traceroute-type app that does incorporate this > function? > > I need a quick app (nothing that generates a significant amount of network > traffic) --where I can enter one command in terminal which will display a > printout of a traceroute, to show the ISP or to show the tech in real-time > when he comes out. The one I send the ISP needs the time/date in order to be > effectively convincing to them. We're supposed to get 20,000 kb/s (up) here, > and quite often, we're not, even with very low internet use in the PCs I'm > running here. Even the download speed's choking... seems like something big > right close is often are clogging the pipe. Traceroute I'm doing is only > going from here to Cambridge. > > I did install a couple of other traceroute-type apps (such as tcptraceroute) > but their man pages didn't seem to show this capability either. I liked > tcptraceroute because it has a "don't fragment" option. Would like to > time/date-stamp that. > > Thanks, > Liz J > > _______________________________________________ > Wlug mailing list > Wlug(a)mail.wlug.org > http://mail.wlug.org/mailman/listinfo/wlug > >

June 6, 2009

Time and Date Stamp on Traceroute?

by E Johnson

Does anyone know a quick command I can use to add a time and date stamp to traceroute? Man page for traceroute doesn't seem to have that option. Or is there a similar traceroute-type app that does incorporate this function? I need a quick app (nothing that generates a significant amount of network traffic) --where I can enter one command in terminal which will display a printout of a traceroute, to show the ISP or to show the tech in real-time when he comes out. The one I send the ISP needs the time/date in order to be effectively convincing to them. We're supposed to get 20,000 kb/s (up) here, and quite often, we're not, even with very low internet use in the PCs I'm running here. Even the download speed's choking... seems like something big right close is often are clogging the pipe. Traceroute I'm doing is only going from here to Cambridge. I did install a couple of other traceroute-type apps (such as tcptraceroute) but their man pages didn't seem to show this capability either. I liked tcptraceroute because it has a "don't fragment" option. Would like to time/date-stamp that. Thanks, Liz J

June 6, 2009

Re: [Wlug] Log Rotation (Snapshot)

by Theo Van Dinter

Maybe it's just me, but rsync would probably do the right thing for you unless you have incredibly large logs and/or are already dealing with large amounts of i/o, etc. Otherwise, it's not hard to manually do something like ... $current_log file is $X bytes and $snapshot_log file is $Y bytes, so if $Y > $X just seek($Y) and grab the remainder of the file. The biggest issue is knowing when you need to start over. ie: if the log file is rotated out, then you can't just go by size but may have to validate log contents or whatever. If you write out logs in append-only mode, then the above method works fine. On Sat, May 30, 2009 at 2:39 PM, Tal Cohen <wlug(a)cohen123.com> wrote: > Guys, thanks for all the replies. In the end it looks like I will just end > up doing regular log rotation. > > That said, I still think it is an interesting question: How to maintain > snapshots of logs without both truncating them when taking the snapshot, and > without putting excessive load on the system. Any thoughts? > > > > -----Original Message----- > From: wlug-bounces(a)mail.wlug.org [mailto:wlug-bounces@mail.wlug.org] On > Behalf Of Chuck Anderson > Sent: Saturday, May 30, 2009 10:24 AM > To: wlug(a)mail.wlug.org > Subject: Re: [Wlug] Log Rotation (Snapshot) > > On Sat, May 30, 2009 at 08:40:08AM -0400, Jason Couture wrote: >> Ive always just written a script to append the log file to a file name >> messages.MMDDYY.log and then delete the log file. and then created an >> init script to run it at boot time. That way everything is neatly >> sorted for later viewing, and i can delete the ones i no longer need >> manually. > > logrotate does this automatically: > > /etc/logrotate.conf: > > # see "man logrotate" for details > # rotate log files weekly > weekly > > # keep 4 weeks worth of backlogs > rotate 4 > > # create new (empty) log files after rotating old ones create > > # use date as a suffix of the rotated file dateext > > # uncomment this if you want your log files compressed #compress > > # RPM packages drop log rotation information into this directory include > /etc/logrotate.d > > # no packages own wtmp and btmp -- we'll rotate them here /var/log/wtmp { > monthly > create 0664 root utmp > minsize 1M > rotate 1 > } > > /var/log/btmp { > missingok > monthly > create 0600 root utmp > rotate 1 > } > > # system-specific logs may be also be configured here. > > /etc/logrotate.d/syslog: > > /var/log/messages /var/log/secure /var/log/maillog /var/log/spooler > /var/log/boot.log /var/log/cron { > sharedscripts > postrotate > /bin/kill -HUP `cat /var/run/syslogd.pid 2> /dev/null` 2> /dev/null > || true > endscript > } > > Resulting rotated log files: > > -rw-r-----. 1 root wheel 703684 2009-05-03 03:42 messages-20090503 > -rw-r-----. 1 root wheel 814677 2009-05-10 11:42 messages-20090510 > -rw-r-----. 1 root wheel 779478 2009-05-17 16:06 messages-20090517 > -rw-r-----. 1 root wheel 721265 2009-05-24 15:03 messages-20090524 > -rw-r-----. 1 root wheel 338888 2009-05-30 10:01 messages > _______________________________________________ > Wlug mailing list > Wlug(a)mail.wlug.org > http://mail.wlug.org/mailman/listinfo/wlug > > _______________________________________________ > Wlug mailing list > Wlug(a)mail.wlug.org > http://mail.wlug.org/mailman/listinfo/wlug >

May 31, 2009

Re: [Wlug] Log Rotation (Snapshot)

by J.R. Mauro

Seriously, check out venti. You can delete the logs all you want as long as you do it after the venti backup runs and you'll keep them forever. Venti stores sha1-addressed blocks of data, so multiple copies of the same data take up a smaller amount of space in venti. Also, blocks are LZO compressed for permanent storage. Venti splits the archive into 500MB "arenas" that you can easily burn to CD when full and keep safe. Venti can be read online with several clients with no slowdown, and doing an actual backup (with the vac program) causes virtually no system load increase and only takes a couple of minutes even on things that are appended to a lot (e.g., a maildir that subscribes to LKML). The coolest online-access to a venti server is via vacfs, which will give you a directory tree like this under the mount point: /YYYY/MMDD/path/to/backedup/file You can bind-mount one of the /YYYY/MMDD directories over your current working directory and pretend that it's 2 years ago and access all your files. Only caveat is that it's hard to set up, but I have some scripts. On Sat, May 30, 2009 at 2:39 PM, Tal Cohen <wlug(a)cohen123.com> wrote: > Guys, thanks for all the replies. In the end it looks like I will just end > up doing regular log rotation. > > That said, I still think it is an interesting question: How to maintain > snapshots of logs without both truncating them when taking the snapshot, and > without putting excessive load on the system. Any thoughts? > > > > -----Original Message----- > From: wlug-bounces(a)mail.wlug.org [mailto:wlug-bounces@mail.wlug.org] On > Behalf Of Chuck Anderson > Sent: Saturday, May 30, 2009 10:24 AM > To: wlug(a)mail.wlug.org > Subject: Re: [Wlug] Log Rotation (Snapshot) > > On Sat, May 30, 2009 at 08:40:08AM -0400, Jason Couture wrote: >> Ive always just written a script to append the log file to a file name >> messages.MMDDYY.log and then delete the log file. and then created an >> init script to run it at boot time. That way everything is neatly >> sorted for later viewing, and i can delete the ones i no longer need >> manually. > > logrotate does this automatically: > > /etc/logrotate.conf: > > # see "man logrotate" for details > # rotate log files weekly > weekly > > # keep 4 weeks worth of backlogs > rotate 4 > > # create new (empty) log files after rotating old ones create > > # use date as a suffix of the rotated file dateext > > # uncomment this if you want your log files compressed #compress > > # RPM packages drop log rotation information into this directory include > /etc/logrotate.d > > # no packages own wtmp and btmp -- we'll rotate them here /var/log/wtmp { > monthly > create 0664 root utmp > minsize 1M > rotate 1 > } > > /var/log/btmp { > missingok > monthly > create 0600 root utmp > rotate 1 > } > > # system-specific logs may be also be configured here. > > /etc/logrotate.d/syslog: > > /var/log/messages /var/log/secure /var/log/maillog /var/log/spooler > /var/log/boot.log /var/log/cron { > sharedscripts > postrotate > /bin/kill -HUP `cat /var/run/syslogd.pid 2> /dev/null` 2> /dev/null > || true > endscript > } > > Resulting rotated log files: > > -rw-r-----. 1 root wheel 703684 2009-05-03 03:42 messages-20090503 > -rw-r-----. 1 root wheel 814677 2009-05-10 11:42 messages-20090510 > -rw-r-----. 1 root wheel 779478 2009-05-17 16:06 messages-20090517 > -rw-r-----. 1 root wheel 721265 2009-05-24 15:03 messages-20090524 > -rw-r-----. 1 root wheel 338888 2009-05-30 10:01 messages > _______________________________________________ > Wlug mailing list > Wlug(a)mail.wlug.org > http://mail.wlug.org/mailman/listinfo/wlug > > _______________________________________________ > Wlug mailing list > Wlug(a)mail.wlug.org > http://mail.wlug.org/mailman/listinfo/wlug >

May 30, 2009

Re: [Wlug] Log Rotation (Snapshot)

by Tal Cohen

Guys, thanks for all the replies. In the end it looks like I will just end up doing regular log rotation. That said, I still think it is an interesting question: How to maintain snapshots of logs without both truncating them when taking the snapshot, and without putting excessive load on the system. Any thoughts? -----Original Message----- From: wlug-bounces(a)mail.wlug.org [mailto:wlug-bounces@mail.wlug.org] On Behalf Of Chuck Anderson Sent: Saturday, May 30, 2009 10:24 AM To: wlug(a)mail.wlug.org Subject: Re: [Wlug] Log Rotation (Snapshot) On Sat, May 30, 2009 at 08:40:08AM -0400, Jason Couture wrote: > Ive always just written a script to append the log file to a file name > messages.MMDDYY.log and then delete the log file. and then created an > init script to run it at boot time. That way everything is neatly > sorted for later viewing, and i can delete the ones i no longer need > manually. logrotate does this automatically: /etc/logrotate.conf: # see "man logrotate" for details # rotate log files weekly weekly # keep 4 weeks worth of backlogs rotate 4 # create new (empty) log files after rotating old ones create # use date as a suffix of the rotated file dateext # uncomment this if you want your log files compressed #compress # RPM packages drop log rotation information into this directory include /etc/logrotate.d # no packages own wtmp and btmp -- we'll rotate them here /var/log/wtmp { monthly create 0664 root utmp minsize 1M rotate 1 } /var/log/btmp { missingok monthly create 0600 root utmp rotate 1 } # system-specific logs may be also be configured here. /etc/logrotate.d/syslog: /var/log/messages /var/log/secure /var/log/maillog /var/log/spooler /var/log/boot.log /var/log/cron { sharedscripts postrotate /bin/kill -HUP `cat /var/run/syslogd.pid 2> /dev/null` 2> /dev/null || true endscript } Resulting rotated log files: -rw-r-----. 1 root wheel 703684 2009-05-03 03:42 messages-20090503 -rw-r-----. 1 root wheel 814677 2009-05-10 11:42 messages-20090510 -rw-r-----. 1 root wheel 779478 2009-05-17 16:06 messages-20090517 -rw-r-----. 1 root wheel 721265 2009-05-24 15:03 messages-20090524 -rw-r-----. 1 root wheel 338888 2009-05-30 10:01 messages _______________________________________________ Wlug mailing list Wlug(a)mail.wlug.org http://mail.wlug.org/mailman/listinfo/wlug

May 30, 2009

Re: [Wlug] Log Rotation (Snapshot)

by Chuck Anderson

On Sat, May 30, 2009 at 08:40:08AM -0400, Jason Couture wrote: > Ive always just written a script to append the log file to a file name > messages.MMDDYY.log and then delete the log file. and then created an > init script to run it at boot time. That way everything is neatly > sorted for later viewing, and i can delete the ones i no longer need > manually. logrotate does this automatically: /etc/logrotate.conf: # see "man logrotate" for details # rotate log files weekly weekly # keep 4 weeks worth of backlogs rotate 4 # create new (empty) log files after rotating old ones create # use date as a suffix of the rotated file dateext # uncomment this if you want your log files compressed #compress # RPM packages drop log rotation information into this directory include /etc/logrotate.d # no packages own wtmp and btmp -- we'll rotate them here /var/log/wtmp { monthly create 0664 root utmp minsize 1M rotate 1 } /var/log/btmp { missingok monthly create 0600 root utmp rotate 1 } # system-specific logs may be also be configured here. /etc/logrotate.d/syslog: /var/log/messages /var/log/secure /var/log/maillog /var/log/spooler /var/log/boot.log /var/log/cron { sharedscripts postrotate /bin/kill -HUP `cat /var/run/syslogd.pid 2> /dev/null` 2> /dev/null || true endscript } Resulting rotated log files: -rw-r-----. 1 root wheel 703684 2009-05-03 03:42 messages-20090503 -rw-r-----. 1 root wheel 814677 2009-05-10 11:42 messages-20090510 -rw-r-----. 1 root wheel 779478 2009-05-17 16:06 messages-20090517 -rw-r-----. 1 root wheel 721265 2009-05-24 15:03 messages-20090524 -rw-r-----. 1 root wheel 338888 2009-05-30 10:01 messages

May 30, 2009

Re: [Wlug] Log Rotation (Snapshot)

by Jason Couture

Ive always just written a script to append the log file to a file name messages.MMDDYY.log and then delete the log file. and then created an init script to run it at boot time. That way everything is neatly sorted for later viewing, and i can delete the ones i no longer need manually. On Tue, May 26, 2009 at 10:56 AM, Gregory Boyce <gregory.boyce(a)gmail.com> wrote: > On May 26, 2009, at 9:31 AM, "Tal Cohen" <wlug(a)cohen123.com> wrote: > > I'm trying to setup a secondary log rotation mechanism on on of my > production systems. > > Right now, the system is setup to rotate logs on a weekly basis. > > In addition to the weekly log rotation, I also need to take hourly > snapshots (i.e. just the deltas between snapshots). > > Is the weekly rotation really needed, or do you just need to retain a > certain amount of data? > Switching to hourly rotation and retaining a weeks worth or more of data > would be the simplest solution. > > As part of this process, I can not delete the original file since that is > being handled by the weekly log rotation and archival process. > > These files get pretty big, so IO and CPU are both concerns. > > Do you have plenty of space? If you do you can have syslog log the data > into two files, one with a weekly rotation and one daily. > -- > Greg > _______________________________________________ > Wlug mailing list > Wlug(a)mail.wlug.org > http://mail.wlug.org/mailman/listinfo/wlug > >

May 30, 2009