stompro.org

At work I have an old Dell SC1435 server that I've been using as a Proxmox server.  It has always been a dog when it comes to IO performance.  It has a basic dell raid card, but it has no battery backed ram, so the write speed has always been slow, and it seems like the raid card is using sata I (1.5Gb/s ) links since I can never get over 140MB/s from it.  The card is a LSI Logic / Symbios Logic SAS1068 PCI-X Fusion-MPT SAS -SAS5/iR . 
 

Supermicro SYS-5018A-FTN4 A1SRi - Intel Atom C2758
I'm testing out a couple of these servers for use in remote sites as crashplan Proe servers + Content Filter (squid + dansguardian) + Misc other uses.  I've also used several of these boxes from Netgate as pfsense firewalls (C2759 Model).
 
 

Memory
2 sticks of Kingston Technology ValueRAM 8GB 1600MHz DDR3L PC3-12800 ECC CL11 1.35V SODIMM Notebook Memory KVR16LSE11/8

Memtest86 4.3.7 Results

I just spent a few hours with a Ricoh tech finding out that no one at Ricoh knows what ports need to be open for their remote fax feature to work over a VPN, through a firewall.  They are going to send the question to Japan for the engineers to work on it eventually. 

So I'll do a packet capture and run it though wireshark to figure out and document what ports are needed.

Update 6/10/14 - Ricoh support figured it out, the ports that need to be open for the remote fax are the range of TCP 54800- 58599.

For some reason my Zabbix history_uint table has grown to 25GB and 293 million rows.  The zabbix housekeeping process isn't keeping up, which is understandable since it deletes rows one at a time.  The history_uint table has only one index on the combination of (itemid,clock) so if you try to delete based on those separately (like delete all rows older than a certain timestamp), you end up with a sequential scan of all 293 million rows.  To get around this we delete things in batch, using a temp table to hold each batch.

Notes about stuff I always add to my Debian installs.

I use a bunch of external USB hard drives for imaging at work, and I've finally figured out how to steamline the process.  Clonezilla is a great tool, but it gets annoying to go through the same exact steps when you are imaging 20 machines at a remote location.  If I had more bandwidth I would look into setting up a clonezilla server, but for right now doing a external USB hard disk to internal disk clone in 20 minutes works pretty good.

Here are some extra UserParameters that I have found useful to add to my Debian systems for Zabbix Monitoring.

The first allows the agent to return the version of Debian that the system is currently running, so if you want to see what all your systems are running at once.  It just cats the contents of /etc/debian_version.