Hello,

 

I've been running unraid for 3+ years.  I recently upgraded to 4.5.6.  Last night I couldn't access my media and it took some time to figure out it was the server.  I had lost connection with it.  I couldn't telnet in, no ping response, no webGUI, nothing on the connected screen.

 

I ended up doing a hard restart.  It booted up okay, but there was no syslog for that session.  It automatically went into a parity check. 

 

I left it running (takes approx 21 hours) and this morning woke to find I have lost the connection again - no access to it. The attached monitor says no signal input.  It doesn't sound like the parity check is still running - which it should be) .

 

What should I do now?

 

Can I stop the auto parity check once it starts?  How?  I carried out a manual parity check only a week or so ago and there were no problems.

 

Thx.

 

 

Happened the same thin to me twice in 3 days. In my case the parity check did end with 5 syncs errors in the first time and 0 in the second time.

 

Im also running 4.5.6

 

Weird thing,I completely lost connection with the box. I never could figure out why.

Happened the same thin to me twice in 3 days. In my case the parity check did end with 5 syncs errors in the first time and 0 in the second time.

 

Im also running 4.5.6

 

Weird thing,I completely lost connection with the box. I never could figure out why.

 

This is a first for me. Been using 4.3.3 for very long time without issue. I'm wondering if there is an issue with 4.5.6 (perhaps some hardware I have)??

 

I'd like to know how to stop the auto parity check once it starts - especially as the last manual one I carried out worked.

 

Also, as the router sets the ip address, I often don't know what it is. I'd like to now set a static one so I always know what it is.  Thx for the comment.

I've been using JoeL's excellent "cache_dirs" program under unMenu.  I was wondering if this is using a lot of RAM - along with the server doing something that also uses a lot of RAM could possibly make it hang?  I have 2gb RAM.

 

I've tried pushing the power button one click as read in another post, along with resetting the router. I also reseated the network cable after unplugging it. Still no joy with a connection. 

 

I guess another hard reset is the only option at this point?? I was going to leave it until it passes the 21 hour mark (the time a full parity check takes) to see if the server comes back on line.  As I said, I'm not sure it is currently doing anything, as it usually makes a slightly different sound when it's accessing the discs.

I've been using JoeL's excellent "cache_dirs" program under unMenu.  I was wondering if this is using a lot of RAM - along with the server doing something that also uses a lot of RAM could possibly make it hang?  I have 2gb RAM.

Absolutely it can run out of memory.

 

Think of it this way viewing a movie will read anywhere from 1Gig to 35 Gig of disk.  That would take anywhere from 1/2 of your RAM to over 10 times more memory than you have if you attempted to buffer it all in memory.

 

Cache_dirs attempts to buffer all it can.  Unless you use the  "-p NNN" option to lighten up the cache pressure to keep everything in memory you'll easily run out.   The linux OS will kill off processes in an attempt to get memory for processes requesting it. The first to go are those idle the longest.  emhttp (the unRAID web-interface) and smtp (the SAMBA daemon process providing shared folders on the LAN) are usually the first to get killed off.

 

For my personal server I need to exclude my entire "data" share and "Pictures" share to have cache_dirs work without running me out of memory. (I do it so we can browse Movies without delays)

 

I would suggest

cache_dirs -p 200

if you think you are running out of memory.

I've been using JoeL's excellent "cache_dirs" program under unMenu.  I was wondering if this is using a lot of RAM - along with the server doing something that also uses a lot of RAM could possibly make it hang?  I have 2gb RAM.

Absolutely it can run out of memory.

 

Think of it this way viewing a movie will read anywhere from 1Gig to 35 Gig of disk.  That would take anywhere from 1/2 of your RAM to over 10 times more memory than you have if you attempted to buffer it all in memory.

 

Cache_dirs attempts to buffer all it can.  Unless you use the  "-p NNN" option to lighten up the cache pressure to keep everything in memory you'll easily run out.   The linux OS will kill off processes in an attempt to get memory for processes requesting it. The first to go are those idle the longest.  emhttp (the unRAID web-interface) and smtp (the SAMBA daemon process providing shared folders on the LAN) are usually the first to get killed off.

 

For my personal server I need to exclude my entire "data" share and "Pictures" share to have cache_dirs work without running me out of memory. (I do it so we can browse Movies without delays)

 

I would suggest

cache_dirs -p 200

if you think you are running out of memory.

 

Thx Joe - really appreciate the insight.  I have been using this app for several months without issue.  I only have movie and tvshows on the server and have been able to browse my whole library without disk spin-up. Gave me a big buzz.

 

As I was reflecting on possible causes for the lost connection I realised this could perhaps be one.  As I can't access the server, I'm not really sure what is going on - can't even grab the syslog...so I'm really scratching my head.  Still a couple hours to wait for parity time to be up.  I don't expect it to give my connection back though.

 

BTW - this app works really well and has kept my HDD"s all spun down.  Never had this before so I'm really happy with.  Thx for your effort here and sharing it with the community.

I've been using JoeL's excellent "cache_dirs" program under unMenu.  I was wondering if this is using a lot of RAM - along with the server doing something that also uses a lot of RAM could possibly make it hang?  I have 2gb RAM.

Absolutely it can run out of memory.

 

Think of it this way viewing a movie will read anywhere from 1Gig to 35 Gig of disk.  That would take anywhere from 1/2 of your RAM to over 10 times more memory than you have if you attempted to buffer it all in memory.

 

Cache_dirs attempts to buffer all it can.  Unless you use the  "-p NNN" option to lighten up the cache pressure to keep everything in memory you'll easily run out.   The linux OS will kill off processes in an attempt to get memory for processes requesting it. The first to go are those idle the longest.  emhttp (the unRAID web-interface) and smtp (the SAMBA daemon process providing shared folders on the LAN) are usually the first to get killed off.

 

For my personal server I need to exclude my entire "data" share and "Pictures" share to have cache_dirs work without running me out of memory. (I do it so we can browse Movies without delays)

 

I would suggest

cache_dirs -p 200

if you think you are running out of memory.

 

Thx Joe - really appreciate the insight.  I have been using this app for several months without issue.  I only have movie and tvshows on the server and have been able to browse my whole library without disk spin-up. Gave me a big buzz.

 

As I was reflecting on possible causes for the lost connection I realised this could perhaps be one.  As I can't access the server, I'm not really sure what is going on - can't even grab the syslog...so I'm really scratching my head.  Still a couple hours to wait for parity time to be up.  I don't expect it to give my connection back though.

 

BTW - this app works really well and has kept my HDD"s all spun down.  Never had this before so I'm really happy with.  Thx for your effort here and sharing it with the community.

Can you log in on the system console?

 

If so, you can save a copy of the system log.  Instructions under troubleshooting in the wiki.

 

If it is a memory issue (running out) it is possible to re-start the SAMBA processes that provide shared directories on the LAN by typing

/root/samba restart

after logging in as "root" on the system console.

 

Now, unless you changes the cache_pressure and free up some memory it may just get killed off again.

 

To change the cache pressure you can type:

sysctl  vm.vfs_cache_pressure=200

 

To flush the caches and free up the memory they were using type:

sync; echo 3 > /proc/sys/vm/drop_caches

I've been using JoeL's excellent "cache_dirs" program under unMenu.  I was wondering if this is using a lot of RAM - along with the server doing something that also uses a lot of RAM could possibly make it hang?  I have 2gb RAM.

Absolutely it can run out of memory.

 

Think of it this way viewing a movie will read anywhere from 1Gig to 35 Gig of disk.  That would take anywhere from 1/2 of your RAM to over 10 times more memory than you have if you attempted to buffer it all in memory.

 

Cache_dirs attempts to buffer all it can.  Unless you use the  "-p NNN" option to lighten up the cache pressure to keep everything in memory you'll easily run out.   The linux OS will kill off processes in an attempt to get memory for processes requesting it. The first to go are those idle the longest.  emhttp (the unRAID web-interface) and smtp (the SAMBA daemon process providing shared folders on the LAN) are usually the first to get killed off.

 

For my personal server I need to exclude my entire "data" share and "Pictures" share to have cache_dirs work without running me out of memory. (I do it so we can browse Movies without delays)

 

I would suggest

cache_dirs -p 200

if you think you are running out of memory.

 

Thx Joe - really appreciate the insight.  I have been using this app for several months without issue.  I only have movie and tvshows on the server and have been able to browse my whole library without disk spin-up. Gave me a big buzz.

 

As I was reflecting on possible causes for the lost connection I realised this could perhaps be one.  As I can't access the server, I'm not really sure what is going on - can't even grab the syslog...so I'm really scratching my head.  Still a couple hours to wait for parity time to be up.  I don't expect it to give my connection back though.

 

BTW - this app works really well and has kept my HDD"s all spun down.  Never had this before so I'm really happy with.  Thx for your effort here and sharing it with the community.

Can you log in on the system console?

 

If so, you can save a copy of the system log.  Instructions under troubleshooting in the wiki.

 

If it is a memory issue (running out) it is possible to re-start the SAMBA processes that provide shared directories on the LAN by typing

/root/samba restart

after logging in as "root" on the system console.

 

Now, unless you changes the cache_pressure and free up some memory it may just get killed off again.

 

To change the cache pressure you can type:

sysctl  vm.vfs_cache_pressure=200

 

To flush the caches and free up the memory they were using type:

sync; echo 3 > /proc/sys/vm/drop_caches

 

Hi Joe,

 

Can't log in via telnet or putty.  Seems a dead stick.  I'm about to do a hard reset. Hopefully it's saved a log?

 

I'm suspicious about the auto-parity check causing issues.  Can you stop it from running once it's started? I'd like to stop it, then activate a manual check if possible.