Warning: fopen(/tmp/config.lock): failed to open stream

Derelict · Dec 29, 2014, 8:16 AM

All modern intel processors are amd64, bro. AMD won that one. Install the 64-bit version. https://forum.pfsense.org/index.php?topic=84679.0

I don't think the file systems are full. I think something on a failing disk or a weird install case is reporting a disk full.

phil.davis · Dec 29, 2014, 1:42 PM

The console screen picture shows disk errors. So I expect the disk is not really full, the disk writes or reads are failing.
If it was running nicely for some time and the problem has only now started, then I guess a failing disk.
If the problem has always been there, then some issue between the hardware and the FreeBSD disk drivers.

stephenw10 · Dec 29, 2014, 3:12 PM

The picture here is getting muddied by multiple reports across different systems etc.
@tarun:

I have just concerned with one of my colleague, he has installed AMD64 but faced the same issue.

Is it actually exactly the same symptoms or just similar? Did he try to replicate your setup? Same hardware?

Tarun, you posted one screenshot showing an out of inodes warning. That is potentially a solvable problem. If something in your setup is creating hundreds of files it should be possible to see what is doing it and find out why.

The other errors shown here and on other threads usually point to a failing harddrive. ishtiaqaj has tried multiple hard drives with the same result which then leads me to suspect a bad cable or disk controller or some badly supported hardware. Tarun, have you tried more than one hard drive?

Although the problems here look similar anything that prevents writing to the drive could throw these same errors. The fact that the vast majority of users are running the same version as you and not seeing any errors implies that your install is different somehow. Which Supermicro box are you using? They are well tested though it would be surprising to find something unsupported.

Steve

tarun · Jan 15, 2015, 11:02 AM

Yes, we have tried different hdd, but the issue same. We have face this issue on 3 of our location.

We are using Supermicro 502-2 1U Server 5015A-EHF-D525 w/Intel Atom D525 CPU /4GB RAM/ 32 GB Crucial hdd.

I have just concerned with one of my colleague, he has installed AMD64 but faced the same issue.

Warning: fopen(/tmp/config.lock): failed to open stream: Device not configured in /etc/inc/util.inc on line 123 Warning: flock() expects parameter 1 to be resource, null given in /etc/inc/util.inc on line 134 Warning: fclose(): supplied argument is not a valid stream resource in /etc/inc/util.inc on line 135 Warning: session_start(): open(/var/tmp//sess_fd59264b4e5cf3b6ae1121f1ccf4762a, O_RDWR) failed: No space left on device (28) in /etc/inc/auth.inc on line 1254

stephenw10 · Dec 30, 2014, 1:45 PM

Ok, that's fairly standard hardware, nothing unusual there. Is it identical on all three sites? Are those Crucial SSDs or really old/small hard drives? How old are they? Checking the SMART stats on them would probably be good idea.
If it's none of those then i'd look at something unusual in your config that's common across the sites.

Steve

jimp · Dec 30, 2014, 2:20 PM

@tarun:

I have just concerned with one of my colleague, he has installed AMD64 but faced the same issue.

Warning: fopen(/tmp/config.lock): failed to open stream: Device not configured in /etc/inc/util.inc on line 123 Warning: flock() expects parameter 1 to be resource, null given in /etc/inc/util.inc on line 134 Warning: fclose(): supplied argument is not a valid stream resource in /etc/inc/util.inc on line 135 Warning: session_start(): open(/var/tmp//sess_fd59264b4e5cf3b6ae1121f1ccf4762a, O_RDWR) failed: No space left on device (28) in /etc/inc/auth.inc on line 1254

More likely than not, the disk is dead. If you have multiple units doing it, then it wouldn't be the first time someone got a bad batch of SSDs. We've seen similar bad batches with both Kingston and Crucial in the past.

It comes down to one of three possibilities:
1. The drive really is full, or at least /var and /tmp both are (unlikely)
2. The disk is dead, in which case it's a coin toss as to whether the system will come back after a reboot
3. The disk has stopped responding to the operating system due to a controller issue, which may be OK after a reboot. Sometimes there are firmware updates for SSDs to help with such problems.

More info here:
https://doc.pfsense.org/index.php/Filesystem_Full_/_Out_of_Inode_Errors

tarun · Dec 31, 2014, 10:14 AM

Hi Steve,

Yes it is identical on all three sites,

These Crucial SSDs are brand new hard drives.

Thanks

stephenw10 · Jul 21, 2015, 8:50 AM

Well, as Jim said, they could potentially be from a bad batch of drives if you bought them all together or maybe they have bad firmware that can be updated. :-\ Have you tried using any other drives?

Steve

Iahmad · Jan 1, 2015, 9:23 PM

the same probelm i was facing and it is with SATA harddisk now i used ide harddisk my luck i found 40 gb IDE hard disk and than i chang from bios hard disk detection auto to LBA(Logical Block Addressing) now it is working since two days hope it has solved my probelm.

tarun · Jan 14, 2015, 11:57 AM

Hi Guys,

We have faced the issue again on one of our site and we need to rebuilt the firewal as it was crash after we reboot.

Please let me know if any one has solution for this.

Thanks

stephenw10 · Jan 14, 2015, 1:17 PM

Have you checked the SMART parameters from these SSDs? Firmware updates?

Steve

tarun · Jan 15, 2015, 9:25 AM

Hi Steve,

Yes, we have run SMART parameter on SSD its seems fine and yesterday one more pfsense failed with the same error.

Please have a look on SMART report.

Thanks

Iahmad · Jan 15, 2015, 10:24 AM

the same issue was with me. before i was using sata 500GB and i also check 320GB but now i am using 40GB IDE but first you check from bios go to device information page there press enter by selecting HD chnage the reading method from auto to LBA . this trick work for me now pfsense is working fine since two weeks.

tarun · Jan 15, 2015, 10:47 AM

@ishtiaqaj:

the same issue was with me. before i was using sata 500GB and i also check 320GB but now i am using 40GB IDE but first you check from bios go to device information page there press enter by selecting HD chnage the reading method from auto to LBA . this trick work for me now pfsense is working fine since two weeks.

ishtiaqaj, we are not using IDE hard disk, we are using SSD hard disks and for few weeks our firewall is also working fine but its gives error in between after few weeks.

Iahmad · Jan 16, 2015, 4:47 PM

@tarun:

@ishtiaqaj:

the same issue was with me. before i was using sata 500GB and i also check 320GB but now i am using 40GB IDE but first you check from bios go to device information page there press enter by selecting HD chnage the reading method from auto to LBA . this trick work for me now pfsense is working fine since two weeks.

ishtiaqaj, we are not using IDE hard disk, we are using SSD hard disks and for few weeks our firewall is also working fine but its gives error in between after few weeks.

dear tarun, you should also check bios hd option there are three option lbs chs and large you should try large mai b it solve your probelm. yes the firewall work someday and than down…

stephenw10 · Jan 17, 2015, 11:18 PM

The smart parameters all look reasonable.
You said you tried multiple drives but were they all identical SSDs? As Jim suggested here and Chris in your other thread it's not unheard of for a whole batch of drives to have firmware bug that your particular setup is triggering.
Have you tried running from a standard SATA drive as a test?

Steve

keshkesh · Jul 21, 2015, 7:20 AM

I have encountered the same problem, when I activated

System: Advanced: Miscellaneous
Use RAM Disks with default /tmp (40 mb) /var (60 mb) ram disk sizes and
with 3hrs periodic frequencies

I had to restore configuration to an earlier state without these settings activated via ssh and reboot. (uninstalled havp package & dashboard widget, but I don't think that was the issue)

I can finally log back in via web interface.

here are some info without these settings set up

__________________________________
2.1.5-RELEASE (amd64) 
built on Mon Aug 25 07:44:45 EDT 2014 
FreeBSD 8.3-RELEASE-p16
Intel(R) Core(TM)2 Duo CPU E7400 @ 2.80GHz
2 CPUs: 1 package(s) x 2 core(s)
Memory usage 
18% of 4073 MB
SWAP usage 
0% of 8192 MB
Disk usage
1% of 443G
____________________________
Installed packages:

Cron 0.1.9
Lightsquid 2.41
pfBlocker 1.0.2
Sarg 2.3.6_2 pkg v.0.6.3
squid3 3.1.20 pkg 2.1.2
squidGuard-squid3
1.4_4 pkg v.1.9.12
_________________________________
output of pkg_info

bsdinstaller-2.0.2014.0430 BSD Installer mega-package
compat6x-amd64-6.4.604000.200810_3 A convenience package to install the compat6x libraries
gettext-0.18.1.1    GNU gettext package
libexecinfo-1.1_3   A library for inspecting program's backtrace
libiconv-1.14       A character set conversion library
p5-Digest-HMAC-1.03 Perl5 interface to HMAC Message-Digest Algorithms
p5-IO-Socket-SSL-1.52 Perl5 interface to SSL sockets
p5-Net-SMTP-TLS-0.12_1 An SMTP client supporting TLS and AUTH
p5-Net-SSLeay-1.42  Perl5 interface to SSL
perl-5.12.4_3       Practical Extraction and Report Language
pkg_info: the package info for package 'pkg' is corrupt
pkgconf-0.8.9       Utility to help to configure compiler and linker flags
python27-2.7.3_3    An interpreted object-oriented programming language
samba36-smbclient-3.6.7 Samba "ftp-like" client
talloc-2.0.7        Hierarchical pool based memory allocator
tdb-1.2.9,1         Trivial Database

stephenw10 · Jul 26, 2015, 4:03 PM

There's a good chance the HAVP package could have filled /tmp there. That's normally only applied in embedded (Nano) installs where running HAVP is not recomended.
Increase the size of /tmp if that's the case.

Steve