Hi,
I have OpenBSD3.5-release and i was installing some extra packages through ftp(pkg_add -v ftp://.../packegename) and i have noticied it takes a really long time extracting files(not talking about lots of dependencies). As an example, blackbox took more than 1 hour installing. Is this normal? Are the packages very compressed that it requires many time do decompress?

I have a AMDK6-II 333Mhz, 164Mb ram.

Thanks.

How fast is your connection? The files aren't compresses that much.

Next time you run it you might be interested in running:
vmstat -w 5 > vmstat.log &; netstat -w 5 >netstat.log&

In the background. This will show how your network is being used and your system resources. I know you might not be able to read them (or maybe you can -- they both have fairly good manual pages and documentation) but if you are stuck we might be able to help.

Note: Please don't post a whole hours worth of these logs here. If you think you need to post them... ask us first so we can tell you how much we need from each log and where in the log file to find it.

I have a broadband cable connection so it shouldn't be a problem(i installed openbsd via ftp and it was quite fast).
I don't really have knowledge at this moment to analise the data so i would really appreciate some help. What part of the output is important?
I really don't understand why this happens. It is on the 3rd hour doing xmms-1.2.7 instalation.

This is a sample of what vmstat will spit out... with just one line. If any of the first three things are non-zero then you might have a bottleneck.
r in run queue
b blocked for resources (i/o, paging, etc.)
w runnable or short sleeper (< 20 secs) but swapped

Most likely, on a slower system you will see it in the first on (r for run queue). If you are having a bunch of conflicts trying to get some resource it will be in the second one.

Also of interest is the last three numbers:
us user time for normal and low priority processes
sy system time
id cpu idle

If idle is 0%... (or 1-2%) then you know what the problem is. Your system is working as hard as it can to unpack the file but it is just taking a long time. If you still have 40 or 50% idle then something is going on. And we can look at the rest of the numbers.

Ok, have checked both the netstat and vmstat output and what matters of it is vmstat--> procs memory page disks traps cpu
r b w avm fre flt re pi po fr sr wd0 wd1 int sys cs us sy id
0 2 0 12840 125944 2 0 0 0 0 0 0 0 230 8 5 0 0 100
From what i understand I have some kind of conflict. Can someone please explain what exactly it is and how to solve it_
Thanks.

PS-this line represents the average of all the vmstat output during the 4, 5 hour that took mozilla/firefox to install-not completly.

Oh I dunno mates. Running FreeBSD on PII 225Mhz ADSL connection and it takes a while to add a large package. Maybe try without the -v ???

How long is a while? Because mine took more than 4 hours to install mozilla-firefox-0.8.tgz on a clean openbsd3.5 install(with idle between 95%-100%, so...).
I have tried without the -v option.
I would really like some feedback from someone that knows how to analyze the vmstat output.

Thanks for all the reaplies until now. This is really letting me down on BSD, but i'm trying not to give up, because i think it's worthed.

According to vmstat it seems that your disk is what is causing the slowdown. That 2 in the second column says that two processes are waiting for some sort of input or output (which is usually disk). The fact that this is constantly on says that something is wrong.

I find this really odd because it doesn't look like EITHER of your disks is being accessed... which sort-of blows my understanding out of the water. None of the other numbers look bad at all. Hmm, I am going to RTM for a little and see if anything sticks out. But there is something seriously wrong here... for some reason your pkg_add program blocks and stays blocked for an inordinate amount of time... without apparent reason.

Ok, so i have a bottleneck.
bottleneck -- Occurs when demand for a particular resource is beyond the capacity of that resource and this adversely affects other resources. For example, a system has a disk bottleneck if it is unable to use all of its CPU power because processes are blocked waiting for disk access.

In practice how do I solve this or what can make this?

That is what I am trying to figure out. BTW, do you have SCSI disks or IDE ones? Because that display shows ide and ... well to be honest I don't know.

If you have two disks one thing you can do is split the load across them (by putting commonly accessed things on different disks). But I don't see any requests like that. Normally you would have one disk with like 50 accesses and the other with none... and that would give you an idea of what to do.

I am not so sure it is your disks... because the slowdown is WAY too dramatic for that and it doesn't look like your disks are being used at all.

Assuming you can figure out what is causing the bottle neck, you can work around it -- like moving partitions -- or upgrade that one part. But don't worry about upgrading yet... because I don't think it is your disk.

We might have to create a report of what is going on, with all the information we have, and send it to the OpenBSD developers if we can't think of anything else. I don't know if it is hardware on your end.

Well, i have 2 ide disks and the master has windows and the slave has OpenBSD.
I haven't really changed nothing and it worked great when i had linux(on slave), so it shouldn't be hardware problem, but I really don't know.

I suspected something similar to that. It sounds like something funky is going on.

You should collect some information and send it to the OpenBSD list... maybe this is an issue they know about or one that should.

lo0 in lo0 out total in total out
packets errs packets errs colls packets errs packets errs colls
94 0 94 0 0 17744 0 14039 0 0
0 0 0 0 0 15 0 1 0 0
0 0 0 0 0 17 0 4 0 0
0 0 0 0 0 15 0 5 0 0
0 0 0 0 0 13 0 2 0 0
0 0 0 0 0 10 0 0 0 0
0 0 0 0 0 13 0 2 0 0
0 0 0 0 0 12 0 2 0 0
0 0 0 0 0 11 0 3 0 0
0 0 0 0 0 14 0 4 0 0
0 0 0 0 0 10 0 0 0 0
0 0 0 0 0 14 0 2 0 0
0 0 0 0 0 12 0 0 0 0
0 0 0 0 0 13 0 3 0 0
0 0 0 0 0 14 0 3 0 0
0 0 0 0 0 12 0 2 0 0
0 0 0 0 0 14 0 3 0 0
0 0 0 0 0 9 0 1 0 0
0 0 0 0 0 16 0 4 0 0
0 0 0 0 0 10 0 0 0 0
0 0 0 0 0 11 0 2 0 0
Just though about putting how my netstat -w 5 looks like when installing.
About the information to send to the mailing lists, what information would be relevant in this case?

Well, you will want to include a full dmesg. A full uname, a copy of your fstab, and examples of your vmstat -w 5 while installing a package. You also want an example of your netstat but with the external interface. lo0 is the loopback one and won't tell them anything.

netstat -I ep0 -w 5 --- replace ep with whatever your external connection is through.

How to get these:

dmesg > dmesg.out
uname -a > uname.out
cp /etc/fstab fstab
vmstat -w 5 > vmstat.out &
netstat -I [YOUR DEVICE] -w 5 > netstat.out &
pkg_add [whatever package you want to try as an example.]

When that is done you will want to trim the vmstat.out and netstat.out files down to about 50 lines apiece. Then you would collect all of this. Write up a description of what is happening, what you have ckecked out, and include a descriptive subject line.

Then you would send it to the appropriate list. I am not on any of the OpenBSD lists but I think the misc list is probably the best starting place. They will direct you to another list if it isn't. Be sure to join the list as most responses will be sent back to the list itself and may not be directly sent to you. The instructions for using the lists are here:
http://openbsd.org/mail.html

That page will cover most of what I just said and will give you an idea of how to format your message. Be sure to include all the information you gather. Too much information is rarely a problem.