ArchStats: Beta testers wanted

farphel · 2003-10-17 23:21:05

Hi All,

Ok, I've got the initial release of 'ArchStats' complete and ready for testing. The ArchStats project was proposed here http://bbs.archlinux.org/viewtopic.php?t=1463 a few weeks back.

I'm looking for people willing to help test it out by downloading the client program and following the instructions for configuring it and running it. You will need to have the python scripting language installed on your local system for the client program.

For more information (and the client program), please visit the beta test site: http://www.coding-zone.com/archstats.phtml. All of the source code is available for review if you are concerned about privacy/security issues.

Thanks,
farphel

terrapin · 2003-10-17 23:53:03

I installed and ran archstats on my play machine. Everything worked as advertised. I hope everything reported correctly. I got machine 3

kritoke · 2003-10-18 00:35:51

I just ran it on my machine, I think I am number 6. It is kind of cool looking at what everyone has on their machines and such. I hope more people use this thing, it could really help out in what things should be added to the repositories, like now we will know how many people use packages from incoming.

Kritoke

farphel · 2003-10-18 00:36:41

First bug found and fixed. The 'longest uptime' was getting overwritten for every system when a client was updating their stats.

This was a fix on the 'server' code which will generate a new server md5. Existing clients should receive a warning message, but can ignore it.

contrasutra · 2003-10-18 01:27:30

uploaded mine.

It said the most popular kernel was 2.6.0-test7-mm1(my kernel). Thats odd, unless I hold more power than others. :?

Also, thank you for making it in Python, it will be fun learning from the script.

apeiro · 2003-10-18 03:15:16

Very cool, farphel! this will be a handy resource.

And here's my first bug report!

After running archstats -i, I tried the --dummy option:

[jvinet@saturn jvinet]$ bin/archstats --dummy
Traceback (most recent call last):
  File "bin/archstats", line 923, in ?
    collect_stats_and_update(STATS, CONFIG, CONFIG_PATH, dummy)
  File "bin/archstats", line 395, in collect_stats_and_update
    print_data(stats, config)
NameError: global name 'print_data' is not defined

I can include my .archstats.conf if it helps.

(btw, let us know when you're ready to release this script as an arch package and we'll accomodate)

Jagged · 2003-10-18 04:42:40

another bug for ya... mitchd@aias.no-ip.info is indeed a valid email address, but the script thinks not

ive got machines 9 (aias, my workstation) and 10 (siren, my lan's server)

farphel · 2003-10-18 11:51:24

contrasutra wrote:

It said the most popular kernel was 2.6.0-test7-mm1(my kernel). Thats odd, unless I hold more power than others

Actually it says that yours is the 'highest' kernel version to date.

Thanks for the help testing. I'll be working on the bugs later tonight and will hopefully have a new client up soon. Looking at the stats, I appear to have "YAB" (yet another bug) in the form of displaying system memory. I can't believe that somebody is running Arch with zero RAM AND zero swap!! :shock:

Just a note about updates to the code: Each time I make changes to the client, I'll need to update the server to recognize the new client's version/md5 hash. Making changes to the server code will cause it's md5 hash to change as well. When you run your client, it will warn you that the server's md5 hash (or version) has changed. In most cases, you can ignore this, but it might be a good idea (from a security standpoint) to pop over to the ArchStats project page and make sure the version/md5 match what your client received. Of course this 'security' measure of comparing client/server versions/md5 hashes could be forged, but at least it makes it a bit harder for a malicious person to have some fun at my expense...

rasat · 2003-10-18 12:26:19

Uploaded my system info. This is great!
Now I can "spy" if anyone is using hwd and other AMLUG made packages (big smile ).

kritoke · 2003-10-18 13:03:09

I think this will also help you decide what should definately remain or be added to the live cd.

Kritoke

rasat · 2003-10-18 14:52:00

kritoke wrote:

I think this will also help you decide what should definately remain or be added to the live cd.

I am seeing yet a broader picture of ArchStat's utilization to do similar work as a software database I had in mind. With few added MySql queries, the package stats can benefit both developers and common users to a large extent. The beauty with ArchStat, no manual data feeding is required and provides lot of information. Lets wait until farphel finish the basic work. I am sure many interesting ideas will be applied along the upgrades.

http://bbs.archlinux.org/viewtopic.php?t=1251

farphel · 2003-10-18 20:42:11

Ok, I've fixed the current list of bugs and uploaded a new 'archstats' client for download for those of you that are still willing to help test.

Fixed in client:
1) supports the --dummy option (really!)
2) email validation works with 4 character domains

Fixed in server:
1) finds the non-trivial minimum for some stats such as memory/cpu speed/etc.

As I stated above, anytime the code is tweaked in the server code, your 'archstats' program will issue a warning about a mismatch of server version/md5 hash. You can always go to the ArchStats beta test project page to compare the version/md5 with what your client detected.

Thanks again for helping test this out.

Cheers,
farphel

Xentac · 2003-10-20 03:45:29

Think it'd be possible to get actual numbers as well as percentages?

farphel · 2003-10-21 00:18:03

Xentac wrote:

Think it'd be possible to get actual numbers as well as percentages?

Sure! How's that?

MNKyDeth · 2003-10-21 02:11:06

I seem to have a problem with the archstats. Here is my error message. Just so you know I did specify where the pacman file was. It's in the default location as I did not move it. /usr/bin/pacman

Please review your submissions:
Traceback (most recent call last):
File "./archstats", line 1010, in ?
prompt_user(CONFIG)
File "./archstats", line 403, in prompt_user
print " Location of pacman: '%s'" % config['PACMAN_PATH']
KeyError: 'PACMAN_PATH'

If there is something else wrong please let me know or any questions you may have. Thanks.

Sorry just found my problem. The md5sum is bad on my copy. I tried to DL it several more times but it still does not match what is on the archstats site. Here is the md5sum I get.

13b8a7e2abb4a090ae59a898fb65d894 archstats

It never chages so I think I am DL'ing it properly. I have tried right clicking on the link and doing a save as and also opening the page and saving the page, but still same md5sum. Maybe I am just slow and did something wrong.

Jagged · 2003-10-21 03:10:07

who has the 99 day uptime?

sarah31 · 2003-10-21 04:02:14

i once had over 240 on my router box

Arielext · 2003-10-21 07:17:14

Oh I will run it on both the Arch boxes I have today

farphel · 2003-10-21 11:45:45

MNKyDeth wrote:

I seem to have a problem with the archstats. Here is my error message. Just so you know I did specify where the pacman file was. It's in the default location as I did not move it. /usr/bin/pacman

You found a bug. I've fixed it and put up a new 'archstats' client (v1.2) 7d4e49e40e10cfa47deaf8a39afaa63a

MNKyDeth wrote:

Sorry just found my problem. The md5sum is bad on my copy. I tried to DL it several more times but it still does not match what is on the archstats site. Here is the md5sum I get.
13b8a7e2abb4a090ae59a898fb65d894 archstats
It never chages so I think I am DL'ing it properly. I have tried right clicking on the link and doing a save as and also opening the page and saving the page, but still same md5sum. Maybe I am just slow and did something wrong.

No, that was the correct md5, it was just the bug causing problems.

However, you pointed out that the ArchStats project page needs to list the valid 'archstats' clients and their respective md5sums so that you can compare after downloading. I've added that to the project page.

Thanks for helping me out with testing! Try downloading the updated archstats client and give it another whirl. You'll need to do the '-i' switch again to make sure it knows where to find pacman.

Cheers,
farphel

MNKyDeth · 2003-10-21 13:43:51

I notice the program has a hard time telling when a system is OC'ed. The system Speedphreak I submitted is a 2500+@2600mhz. It showed up as a reguler Athlon on the list I think. Don't know if it's a bug or if it's just cause I am OC'ed but figured I'd let you know.

farphel · 2003-10-21 14:22:51

MNKyDeth wrote:

I notice the program has a hard time telling when a system is overclocked.

The archstats client just reads from /proc/cpuinfo. If you're overclocking and it's the new speed doesn't show in /proc/cpuinfo, then I'd say that the kernel doesn't know that your CPU is overclocked. I'm sure it still runs at the higher speed, it's just that the kernel doesn't know/care about it.

Anyway, as a result, your stats will only reflect what is in /proc.

hth,
farphel

Jagged · 2003-10-21 14:48:48

my OC'ed celeron shows just fine.. it's a 333 mhz chip, but it shows it /proc/cpuinfo as 501.4 mhz

farphel · 2003-10-21 15:57:46

Jagged wrote:

my OC'ed celeron shows just fine.. it's a 333 mhz chip, but it shows it /proc/cpuinfo as 501.4 mhz

That's cool! Do you think that all OC'd chips reflect their enhanced speed in /proc/cpuinfo, or is it a hit-or-miss kind of thing? Maybe if something that can be set via the BIOS it would be detectable, but I'm really not a hardware person.

Any thoughts as to why MNKyDeth's OC'd stats don't show in /proc/cpuinfo?

Xentac · 2003-10-21 16:02:41

farphel wrote:

Xentac wrote:
Think it'd be possible to get actual numbers as well as percentages?
Sure! How's that?

That's sweet man. Thanks.

terrapin · 2003-10-21 16:33:43

farphel,

Quick question. What happens when someone removes a package from their system? Say someone has "unique package" on thier system and they are the only one that has it. If they remove "unique package" will it be deleted from the database? Also what happens if packages are upgraded like from gnome 2.2 to gnome 2.4. Will the old package be removed or will there be two entries in the database one for gnome 2.2 and one for gnome 2.4?

Arch Linux

#1 2003-10-17 23:21:05

ArchStats: Beta testers wanted

#2 2003-10-17 23:53:03

Re: ArchStats: Beta testers wanted

#3 2003-10-18 00:35:51

Re: ArchStats: Beta testers wanted

#4 2003-10-18 00:36:41

Re: ArchStats: Beta testers wanted

#5 2003-10-18 01:27:30

Re: ArchStats: Beta testers wanted

#6 2003-10-18 03:15:16

Re: ArchStats: Beta testers wanted

#7 2003-10-18 04:42:40

Re: ArchStats: Beta testers wanted

#8 2003-10-18 11:51:24

Re: ArchStats: Beta testers wanted

#9 2003-10-18 12:26:19

Re: ArchStats: Beta testers wanted

#10 2003-10-18 13:03:09

Re: ArchStats: Beta testers wanted

#11 2003-10-18 14:52:00

Re: ArchStats: Beta testers wanted

#12 2003-10-18 20:42:11

Re: ArchStats: Beta testers wanted

#13 2003-10-20 03:45:29

Re: ArchStats: Beta testers wanted

#14 2003-10-21 00:18:03

Re: ArchStats: Beta testers wanted

#15 2003-10-21 02:11:06

Re: ArchStats: Beta testers wanted

#16 2003-10-21 03:10:07

Re: ArchStats: Beta testers wanted

#17 2003-10-21 04:02:14

Re: ArchStats: Beta testers wanted

#18 2003-10-21 07:17:14

Re: ArchStats: Beta testers wanted

#19 2003-10-21 11:45:45

Re: ArchStats: Beta testers wanted

#20 2003-10-21 13:43:51

Re: ArchStats: Beta testers wanted

#21 2003-10-21 14:22:51

Re: ArchStats: Beta testers wanted

#22 2003-10-21 14:48:48

Re: ArchStats: Beta testers wanted

#23 2003-10-21 15:57:46

Re: ArchStats: Beta testers wanted

#24 2003-10-21 16:02:41

Re: ArchStats: Beta testers wanted

#25 2003-10-21 16:33:43

Re: ArchStats: Beta testers wanted

Board footer