Are you using a mac PC. If not you don’t need netatalk. If you are not using nfs mounted file system. You don’t need nfs.
Winbind is a program that allows users in a heterogeneous network to log in using workstations that have either Unix or Windows NT operating systems. The program makes workstations using Unix functional in NT domains, by making NT appear to look like Unix to each Unix workstation.
I have a few macs and a few PCs. I use NFS to stream to my WDTV. I could probably get away with disabling Winbind though, maybe? Iol. I’m not sure, but I could try disabling a few more services. The sleep times are actually much better than they were (7 second wake up was ridiculous) but they are still pretty bad. Under no normal usage would you be waking up a drive every 30 minutes on a desktop or server, so I’d rather just keep the thing spinning until it dies.
I figure if it’s waking up every 20 minutes or so, it COULD get a little better, but it won’t get to 5 hours without being isolated on it’s own network (because that’s why I purchased a NAS). They should really just disable the sleep function all together and call it a day.
The seven second wake ups seem to be after the system is busy like at 03:00 am. During this busy time
the system cache is flushed. As soon as the system puts the disk to sleep some process needs to access an inode or a code segment that has been flushed. So it goes to disk which wakes it up. After looping thru the code for a few minutes all of the data needed to run has been cached. Now the system does not need to access the disk.
Not sure what’s waking up my disk every 20 to 30 minutes though. I do have
backups taking place on it and media streaming though, so it could be
anything.
It is difficult to determine what is causing the disk to wake up. Here is script I call checkgen1.sh. It sleeps for 4 seconds then checks /proc/diskstats to see is any disk activity has occurred. If it has it then does a find to list those files that have been accessed in the last minute. Since it wakes up every 4 minutes it will display the same file names during that minute. You just need to run it for a couple of minutes if the system is not sleeping. If the system is sleeping you can let it run a little longer. If you are using putty turn on logging. So that it will save the output.
file checkgen1.sh
#!/bin/bash
function test () {
if [ “$3” != “$2” ]; then
da=date +%k-%M-%S
A=$2
B=1
ior_sda=${A#-}
iow_sda=${A%-}
A=$3
ior_sdaold=${A#-}
iow_sdaold=${A%-}
let a=ior_sda-ior_sdaold
let b=iow_sda-iow_sdaold
if [ $4 -ne 1 ]; then
echo -n $da " "
fi
printf "%4s %4s %4s " “sda”$1 $a $b
fi
}
function GetData {
for (( i=1; i<8;i++ ));
do
y=printf "%s%s" $3 $i
x=awk -v disk="$y" '{if ($3==disk) printf "%s-%s",$10,$6}' /proc/diskstats
sda[$i]=$x
done
}
while :; do
GetData ior_sda1 iow_sda1 “sda”
B=0
for (( i=1; i<8;i++ ));
do
test $i “${sda[$i]}” “${sdb[$i]}” $B
done
sdb=(“${sda[@]}”)
if [ $B -eq 1 ]; then
echo
find / -path /proc -prune -o -path /sys -prune -o -cmin -1
fi
sleep 4
done
I have this option disabled actually since the beginning - this unit sits in the study - and judging by experience of others I don’t think I’m missing anything here.
Interesting when I run that code script (after fixing I hope the characters the site is stripping out or converting) I get the following entries pretty consistently.
I think the files external_share_size share_size minutes_since_disk_access are used by the monitorio script
pacct_source atop_current is used by atop. I don’t think these files are causing a sleep problem.
The msg.sock file was causing problems in the past and was solved by createing a tmpfs.
Right now I’m leaning towards samba being the problem again. I found on my system that /etc/samba/smbpasswd
and /tmp/browse.dat files are being updated frequently. I’m now testing the system with samba stopped.
Today my system slept 56 second from 11:50am to 5:00pm.
Hi guys,
I’ve just upgraded to the latest V4…320 firmware. Everything seems to work as previously despite the mass of error, warnings, do not match messages, … in the update.log.
Also, the “tmp/dynamicconfig.ini: no such file or directory” is back again. I guess the update or the reboot has removed my manual created file.
I’ll run the sleeptime.sh later on because the system is rebooted since only 2 hours ago and reports 100% wakeup time now.
We’ve been discussing the sleep issues that continue even with this latest firmware in this thread:
It appears, at least for me, the following code that I put in the S98user-start file seems to fix the missing file error messages in the log file(s) for the dynamicconfig.ini file.
