houkouonchi

Here is my ghetto style JBOD chasis: Got free chasis from work and converted it into a JBOD with SAS expander.

I noticed there is a west server now to. I also noticed when I manually chose the size it always appears to go to the texas server. Is that by design? Is there anyway to do a 200MB test on any of the other servers or what size until it forces you to texas?

If you are interested I have a server in L.A. on 1000/1000. It actually also runs speedtest.net's LA server (well half of the time as they seem to balance the load between mine and another). It averages about 35 megabits now (load split) and before was around 70 mbits (day average). L.A. is usually a good option for faster speeds to Asia (until you get something in asia). I tried selecting the DC server but got: :::.. Download Test Results ..::: Download Connection:: 140478 Kbps or 137.2 Mbps Download Test Size:: 200 MB or 204800 kB or 209715200 bytes Download Speed:: 17560 kB/s or 17.1 MB/s Tested At:: http://TestMy.net version:12 Validation Link:: https://testmy.net/db/mnFrcZ3 Test Time:: 2011-10-20 11:32:13 Local Time 1MB Download in 0.06 Seconds - 1GB Download in ~1 Minutes - 2509X faster than 56K This test of exactly 204800 kB took 11.663 seconds to complete Running at 107% of hosts average (houkouonchi.jp) User Agent:: Mozilla/5.0 (X11; U; Linux x86_64; en-US) AppleWebKit/534.7 (KHTML, like Gecko :::.. Upload Test Results ..::: Upload Connection:: 48271 Kbps or 47.1 Mbps Upload Test Size:: 29.2 MB or 29916 kB or 30634200 bytes Upload Speed:: 6034 kB/s or 5.9 MB/s Tested At:: http://TestMy.net version:12 Validation Link:: https://testmy.net/db/CHJ2grL Test Time:: 2011-10-20 11:33:00 Local Time 1MB Upload in 0.17 Seconds - 1GB Upload in ~3 Minutes - 862X faster than 56K This test of exactly 29916 kB took 4.958 seconds to complete Running at 111% of hosts average (houkouonchi.jp) User Agent:: Mozilla/5.0 (X11; U; Linux x86_64; en-US) AppleWebKit/534.7 (KHTML, like Gecko) Chr Almost the exact same results I got to texas. According to iftop my traffic was coming from/going to 174.120.187.140 which is the same IP as this site so I am guessing it did not use the server for some reason.

I also posted in another thread that was super old but you also recently responded in about whether or not you were still interested in having multiple speed test servers?

CABLE are you at all still interested in having multiple servers?

So are you using some sort of whois/IP owned by lookup method now instead of just rDNS? I know my connection is still listed as houkouonchi.jp even though that is just what my rDNS is due to PTR records but my IP is owned by Verizon Communications.

To anyone who was wondering the total storage in that rack (just HD's, not counting what is lost from raid) is 162TB. With colo'd and other machines my total storage is just under 200TB (quite a ways from 1PB). Several of the machines in that rack keep backups of others and do not run 24/7. Only two of the machines run 24/7 (well five I guess if you count the two 30x3TB but those are JBOD chasis and aren't computers). The power usage actually isn't very bad considering everything: I am using the storage mainly for storing media. I just recently added the 90TB and until then my biggest system which contained almost all of my collection was 40TB but I was running out of disk space so I added 90TB. Since I did not just upgrade my current array (and move all my data to it) but instead added additional storage via JBOD chasis My biggest storage machine went from 40TB to 130TB. This storage should last me atleast 3 years. Also since the new 90TB array I added currently has almost nothing on it I have power-management enabled which spins down the drives after 30 minutes which saves a good 120-140 watts of power. I store a lot of high definition TV shows/movies. I also have a ton of Anime and various asian drama's. I don't like to delete anything I download as there are many times I feel like watching something again years later. Also I provide access to my friends/family. My internet connection is good enough (75mbit upload) that I can stream pretty much anything over the internet. It is nice because I can watch anything off my machine at my house when I am at my dad's house (We like to watch some TV shows and movies together). I got a rack because it simply made sense. The high density/high number of hot-swap bay chassis (which I wanted) are rack-mount. I used to just have a rackmount chasis sitting on a dresser or desk like a regular computer for a while. When I bought my house I had several rackmount machines/chasis and I figured I might as well get a rack so I can have things a little better organized. I also spent a little money to convert a few systems into rack-mount. In the end I think it was the right decision.

Its still doing it but I noticed this time its also loading in the random data twice so something weird is going on with the browser I think. Oddly enough it works on the smaller tests.

Working for a web-hosting company so initially that is exactly what I figured it was either a max_execution_time or possibly a fcgi (if using fcgi) timeout. However, even now I am still getting the 500 error on tests over 17 MB and the 29 MB upload test takes only 30 seconds or so to run (if that).

It was about time so I finally updated to firefox 6. The upload test starts *MUCH* faster now (almost as fast as chrome). However; I am still getting an internal server error (500) when testing any of the larger sized tests (over 17MB). So I guess the 500 error was not caused by a timeout.

I am using sabayon. I haven't taken the bullet to upgrade to the much newer versions of firefox (still using the newest 3.6.x) so its possible the newer one will fix this issue. Did you ever see this issue in the past as I have always seen this long delay for as long as I remember using testmy.net on multiple computers running linux/firefox so I thought it was normal. Here is a video of some spedtests I just took now for you. I did speedtest.net testmy.net and VZ's own speedtest applet. It might still be processing though so give it a few minutes.

Reinstalling firefox would not make a difference if it really was plugin related as those are not removed from 'reinstalling' it atleast on linux. Also I have the same problem on a compiled 64-bit version of firefox as well as the 32-bit firefox binary so it surprises me both are affected if it is just a plugin. They are quite a bit different in versions too. Curious if anyone else on linux also sees this problem.

I recently upgraded my storage: Here is my machines and its all for *personal* usage. This rack is in my house: And my connection at home is 150mbits down and 75mbits up =)

It was slow enough to the point that it was causing an internal sever error on the server (I am guessing as it hit a several minute timeout). The delay is during the generating of random data when firefox uses 100% cpu usage on my system (and downloads at a relatively low speed during this time). At first I thought it might just be a server issue until I tested it with chrome and the largest upload test only took a few seconds to start uploading data instead of a few minutes. I was just curious if the slowness is normal or something specific to firefox on linux.

802.11n router is not going to help your laptop get better speeds if its still G. For a laptop I think I would spend the extra time to actually upgrade the internal miniPCI wireless card. This is not very difficult on dells and I don't really like dongle and stuff for my laptop. I suppose if its docked then its not a big deal. Even 802.11n isn't really good for me since i have the 150/75 service. Personally I just use ethernet for pretty much everything. I spent a couple hundred when I bought my house to have ethernet ran (as I didn't want to do it myself).

Some other observations... I noticed that the upstream results are very low (15-20 megabits) vs the 200+ down. If this is a linux server it could be caused by a low rwin setting. Have the settings on the server been tuned on the receiving side? I personally use a web100 kernel on my machine to help with tcp/ip tuning: http://web100.org/ I am also a bit curious what /proc/sys/net/ipv4/tcp_reordering is set to. I have two 35/35 connections that are bonded over VPN at my home but I get very bad results to this site even though the bonding via the VPN is routed through the same network I got 200 megabits to this server. One side-effect of the bonding is the packets are more out of order than usual but I can still get very high speeds to the majority of the servers out there. To this server it was < 1 megabyte/sec but for example to speedtest.net's TX server I got over 60 megabits (7+ megabytes/sec).

The new site is much better. The smart test was broken on fast connections in the past as it would just get stuck on the biggest test and constantly do it over and over again without giving results. I also see that the speed test server is finally on gigabit so I no longer reduce other peoples connections when I do a test from my server: :::.. Download Test Results ..::: Download Connection is:: 206514 Kbps about 206.5 Mbps (tested with 200 MB) Download Speed is:: 25814 kB/s or 25.8 MB/s Tested From:: https://testmy.net/ (Dallas, TX USA) Validation Link:: https://testmy.net/db/P4NMR9VAL Test Time:: 2011-02-09 20:45:51 GMT -7 1MB Download in 0.04 Seconds - 1GB Download in ~1 Minutes - 3601X faster than 56K Tested from a 200 MB file and took 8.124 seconds to complete Running at 121% of hosts average (Houkouonchi.jp) User Agent:: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.1.16) Gecko/20101130 Firefox/3.5.16 [!]

I was testing from a 1 gig connection.

I remember I reported a while back that the speed test get in an infinite loop during speed tests from high speed connections (>70 meg or so). I got an email from this site saying some changes were done and this finally looks like it was fixed as a speed test finally completed: :::.. testmy.net test results ..::: Download Connection is:: 94159 Kbps about 94.16 Mbps (tested with 102400 kB) Download Speed is:: 11494 kB/s Upload Connection is:: 15054 Kbps about 15.1 Mbps (tested with 5983 kB) Upload Speed is:: 1838 kB/s Tested From:: https://testmy.net (Main) Test Time:: 2010-03-31 13:11:49 GMT D-Validation Link:: https://testmy.net/stats/id-OL8XIGU3Q U-Validation Link:: https://testmy.net/stats/id-03C4ZHMOR User Agent:: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.1.7) Gecko/20091221 Firefox/3.5.7 [!] I can't say the upload results were that great and it looks like you guys still don't have any test servers on gigabit?

Well the 25MB test works just fine (as it ramps up each time up to 100MB) but the problem is it is still getting caught on 12 MB where it just keeps repating itself. Unfortunately this prevents me from doing the dual test =(

It didn't do it on the 100 meg test and only did it on the one which ramped up. I am guessing it wouldn't do it on the ones above 25MB either but I will test it again the next time I am at work and remember to test it.

Is there a bug if you are on too fast of a connection? When I tried to do the automatic/dual speed test when it ramps up on the different file sizes it gets stuck on the 12MB one and just keeps reloading the page and using gkrellm I saw a constant download speed of 5-6.5 megabytes/sec basically using a ton of bandwidth on the remote server as leaving it going for just 20 seconds would easily eat up 100+ MB of bandwidth. It doesn't seem to do this on my home connection (using the same laptop) and only does it when I am at work, so does it just get stuck in some sort of loop if you are on a fast connection 40+ megabits? I also tested konqueror out (I was using forefox) and it did the same thing.

Heh, I guess I am the only one so far who picked "50MB+" for download. If you had a 100MB+ I wold have chosen that =P

One/fastest of my home connections (I have 5): :::.. testmy.net test results ..::: Download Connection is:: 24482 Kbps about 24.48 Mbps (tested with 25600 kB) Download Speed is:: 2989 kB/s Upload Connection is:: 1467 Kbps about 1.5 Mbps (tested with 2992 kB) Upload Speed is:: 179 kB/s Tested From:: https://testmy.net (Main) Test Time:: 2008/06/02 - 9:42pm D-Validation Link:: https://testmy.net/stats/id-IDLE1OPS7 U-Validation Link:: https://testmy.net/stats/id-P5RQIYKX8 User Agent:: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.8.1.14) Gecko/20080404 Firefox/2.0.0.14 [!] A much better disk benchmarking method: root@sabayonx86-64: 08:36 PM :~# dd bs=2M count=10000 if=/dev/sdc of=/dev/null 10000+0 records in 10000+0 records out 20971520000 bytes (21 GB) copied, 26.8962 s, 780 MB/s and hdparm.... root@sabayonx86-64: 08:37 PM :~# hdparm -t /dev/sdc /dev/sdc: Timing buffered disk reads: 2208 MB in 3.00 seconds = 735.67 MB/sec

hmm, its accurate on all the computers I have tested, especially when you use the command line client, it defnitely seems accurate, and I use other tools which agree with the speeds like gkrellm, iptraf, etc.