Joel K's questions

I am using an external eSATA device which takes 5 SATA drives and presents them as one large HW RAID5 to linux (ubuntu).

It uses a SiI3726 chipset.

Linux sees and handles the sata device just fine, but I'm wondering if there's any way to query/monitor the RAID5 status that's hiding behind this HW RAID.

(from dmesg -- I swear I don't have a 16T single drive...)

[   16.409678] sd 4:0:0:0: [sde] 31255691264 512-byte logical blocks: (16.0 TB/14.5 TiB)
[   16.409727] sd 4:0:0:0: [sde] Write Protect is off
[   16.409730] sd 4:0:0:0: [sde] Mode Sense: 00 3a 00 00
[   16.409752] sd 4:0:0:0: [sde] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
[   16.463573]  sde: sde1
[   16.463785] sd 4:0:0:0: [sde] Attached SCSI disk

(from lshw)

   *-disk:4
        description: ATA Disk
        product: SATA H/W RAID5
        physical id: 4
        bus info: scsi@4:0.0.0
        logical name: /dev/sde

Any guidance on peering at the system underneath would be appreciated.

RAID status is my big concern, but actual drive models would be helpful too.

I looked at some libatasmart tools, but couldn't find anything regarding SATA port multipliers in general.

There are utilities to check/set the RAID settings in OSX/Windows, so it is apparently possible. I'm looking for a way to do this in Linux and then potentially add to Nagios, etc.

Here's what the HW looks like:

I for one feel that configuring OpenLDAP with LDIFs instead of basic configuration files is a major step in the wrong direction.

Does anyone have some tools/techniques to more easily manage OpenLDAP that don't require LDAP to be working 100%?

I am considering retiring some of our older OpenDJ LDAP servers and replacing with OpenLDAP. (seems to be more widely supported)

I am wondering if it's possible to replicate directly from OpenDJ to OpenLDAP as an interim solution to remove the OpenDJ slaves and then flip the master over later.

Is LDAP replication implementation specific or more general?

I guess I'll have to just go give it a try, but I was looking for advice from someone who's been down this road.

I pay for a DNS service based on queries per second.

We are not using IPv6, but a large number of queries (that I pay for) are coming in for AAAA records. I understand that most DNS stacks will now ask for A and AAAAs at the same time, and that I can't change that.

What I would like to do is put something in the AAAA records with a long TTL. (decreasing my hit rate)

Is there anything I can put? Null? The equivalent IPv4 Address?

Any guidance would be appreciated.

I have two different MySQL instances that I would like to 'slave' to a third instance. (so I can do easy joins on the third instance)

eg.

mysql1> show databases
db1
db2

mysql2> show databases
db3
db4

mysql3> show databases
db1
db2
db3
db4

I've looked at maatkit (percona toolkit) pt-table-sync but people say it can corrupt data. (It apparent removes and re-adds data to generate inserts)

pt-archiver sort of works for 'snapshots', but db1 is approximately 6GB and copying the whole thing over is a lot more data than really needed. The real time updates are only about 100MB a day.

The natural concept to me would be to allow mysql3 to run as a replica slave of BOTH mysql1 and mysql2, but that doesn't seem to be an option in MySQL

Tungsten Replicator seems to allow for this type of data sync, but it seems a bit unwieldy to configure and I'm concerned about reliability.

Does anyone have other solutions they've used for this problem?

Please be doubly sure that you've investigated this on OSX Server. http://www.apple.com/server/macosx/ Which is a different beast from normal OSX.

OSX Server (not the desktop version) includes a built in copy of MySQL. This would probably be OK, but it's missing the development libraries, so you can't do things like compile/install python bindings to work with it.

On OSX desktop I just grab the MySQL binary packages from MySQL's website and python support is built in.

Has anyone found a way to prune out OSX's MySQL installation?

(The reason I need to do this is that only OSX Server can be run inside a VM, and I need to approximate an OSX desktop development environment inside a VM -- XCode, MySQL, etc.)

I've seen people recommend combining all of these in a flow, but they seem to have lots of overlapping features so I'd like to dig in to why you might want to pass through 3 different programs before hitting your actual web server.

nginx:

• ssl: yes
• compress: yes
• cache: yes
• backend pool: yes

varnish:

• ssl: no (stunnel?)
• compress: ?
• cache: yes (primary feature)
• backend pool: yes

haproxy:

• ssl: no (stunnel)
• compress: ?
• cache: no
• backend pool: yes (primary feature)

Is the intent of chaining all of these in front of your main web servers just to gain some of their primary feature benefits?

It seems quite fragile to have so many daemons stream together doing similar things.

What is your deployment and ordering preference and why?

I'm aware of the DOS utility to "Reset" an Intel SSD to factory settings, but I'd rather not shut down my system to do this to my SSDs.

Supposedly newer kernels support TRIM, but I haven't found any utilities to actually run TRIM.

Has anyone found a way to do this?

I'd like to connect two remote sites via a secure point to point tunnel. I've been using OpenVPN, but it's a little flakey and really seems to be more for connecting lots of users rather than a point to point link.

Redhat seems to have some nice built in scripts for building Point to Point IPSEC tunnels, but I haven't found anything similar in Ubuntu.

Any guidance on what to use here and how to do it? I'd like to keep it as simple as possible, but still have basic PSK security.

I'm contemplating setting up a squid (or possibly varnish) cache on a system with SSD drives.

The obvious benefit is that these systems have great READ speeds and I expect my hit ratios to be fairly high.

Let's assume I can put 7 SSDs in to a RAID configuration. (there are some cases that will let me pack in much much more)

Implementation questions:

• Should I use RAID0? (I expect a drive to fail eventually, so this seems dangerous.)

• Should I use RAID10? (This halves my disk footprint, which is costly.)

• Should I use RAID5? (SSDs are known to have "bad" write performance and write limits, and all the extra parity writes may slow this down considerably.)

• Should I just treat each disk as it's own squid datastore? (how well does squid handle multiple data stores? and what happens if/when one fails?)

• Should I ignore datastores and just make the SSDs in to large SWAP partitions and let the linux VM do it's thing? (seems sloppy)

Any advice from folks using SSDs in production environments would be greatly appreciated. (esp if you're using them for HTTP caches)