mister_z

Used to be Dell, now moved to 100% HP. Excellent Linux support, good hardware, breaks less often that Dell has for us. YMMV

For a database that is read-mostly, RAID 5 or 6 is fine. My organization uses it all the time for low write/delete/modify databases. And we use RAID 6 for much of our data warehouses (tempspace and index rebuild areas excepted). RAID 5 is only slower during write ops, not read ops. So consider the usage. RAID 1 and it's big brother RAID 10 we use for fast transactional systems, beginning with the log space (since that's written to with every write op), then for the temp space (large sorts, etc), then as needed for high change rate tablespaces. For spindle capacities > 600GB, we go to RAID 6, since the hotspare rebuild times start to get longer, and we really don't want to lose another drive while one is rebuilding. Else it's "Hope the backup set is current," and "Tell the users it'll be down till the restore finishes." Also, don't confuse throughput/bandwidth with access time. They only correspond when there are enough users active, and then only roughly. Without a demonstrated bottleneck, adding more spindles to a RAID set (thus raising the stripe size) means more users can get to data in the same time, but not necessarily that the same number of users can get to the same data faster. Lastly, remember the hot spares. Our rubric is 1 HS per 30 disks in service if the hot spare can take over for any other spindle in the same VNX or VMAX (my teams manage several data centers with rather sizeable SANs in each). If you do not have hotspares, you MUST HAVE ACTIVE MONITORING watching for failures. I have seen too many organizations lose a disk in a redundant system and not know it until a second one fails. "Oops, this is going to suck." As I recently explained to a colleague in another division, my preference (not always possible) is to start as small as possible and "watch and grow" the system as performance data becomes available. I advised him to keep 30% of his space in reserve and adjust its placement as he learns more. Best of luck.

Linux Journal Archive 1994-2012 on DVD. ~ $40.00 USD All the back issues, searchable. NOT a textbook, but rather an exploratory journey in a series of bite-sized pieces. Issues focussed on System Admin, Programming, Security, etc. I've been doing Unix a long time, and have done Linux since Slackware, and I still gain nuggets of wisdom from it (I keep in on a USB key with me, and I'm making it a goal to read every article in every issue eventually. Something to do over lunch every day.) Because the truth is, *nix is so big, few of us even have the time to explore all the corners of it, and so there's always something new to learn. Sometimes even in the articles I thought I'd have no interest in. Cheers, Oh, and if you like the archive, consider subscribing to keep up with the newer stuff.

Well, there are many ways to hose Linux -- as root. But if you routinely run things as root, you deserve what you will eventually get. You're basically saying "Yes, I know the gun is loaded, but I want to muck about with it anyway just to see what happens." That's why sudo was invented. If you have to sudo the command, you'd best understand what you are doing. If you are operating as a standard user on a well-managed system (i.e. one with quotas, limits, etc), there's little you can do to screw up anyone save yourself. As root ? You can delete the entire file system, destroy the volume manager information, reformat hard-drives, just about anything.

FWIW, you can get Fedora "spins" with different interfaces, see Fedora Spins, including KDE (my preference), XFCE, etc. And with the base install, you can choose either Gnome 3 or KDE, and tyou can even install the cinnamon variant. Cheers.

Backup: Hmm, depends on standards and the pocket depth. My day job has corporate standards, so NOT using Symantec NetBackup will get people in trouble. So the first question for the interviewer is "Do you have any standard for backup software ? Legato ? TSM ? NetBackup ?" In my pro bono work -- smalll non-profits, etc, for linux installations with more than 1 server, I recommend AMANDA overall. It takes some setup -- but that's why there are administrators. And iits flexible, resilient, and it just works. You can set it up to use common tape devices, including hand-me-downs. Or your can use SATA docks and dump to disks. Combine with LUKS volumes, and you have encrypted backups as well. As for MySQL -- one way is to read-lock the database for consistency and LVM snapshot the volume that the data files reside on. Release the read lock and the database can go back to happily accepting transactions while AMANDA backs up the snapshot volume. In earlier versions of MySQL (5.0 and earlier) this was the only way to get an "almost online" backups in a 24x7 environment. Security procedures ? That can fill a book by itself. And does. The major point about security is this: If it's highly convenient, it's probably insecure. Lock out root ssh logins ? But it's way more convenient to log in as root directly -- why do I have to use sudo ? The DBA team needs root access. And then know what they're doing. I'll just give them the password to save time. Enforcing password changes every 90 days ? What a pain. If I put in password complexity requirements, everyone will be so annoyed at me.\ Creating a cron.allow ? Are you kidding ? Then every time someone wants to run a cron job, I have to add them to the file. Understanding and reluctantly embracing the security mindset is much more important than memorizing every rule. If you embrace security, you can google to build an effective security baselin, and you may even enforce it. Good luck.

First, you need to decide what you are trying to back up. The first step in disaster recovery -- as I evangelize often enough -- is deciding "what disaster am I preparing for ?" (You'd be surprised how often that question gets overlooked. And then you have 4 blind men describing the elephant AFTER many plans have been created.) Anyway, are you trying to be able recover the entire laptop ? You do need some way, unless you want to re-install from scratch, a major pain. I consider this part a job for CloneZilla. Periodically. And a USB drive is an excellent place to put it. Plug in the USB drive, boot from a thumb drive, answer a few questions, let it chug. But you will want to protect daily work as well -- say her home directory. Are you okay with working at the command line ? For my wife's laptop, I use rsnapshot to dump it to my back-end server. Especially nice since it allows her to recover earlier versions. of certain documents. see rsnapshot for details. Note that while the setup part -- done only once -- takes some ability to work at the command line and to closely follow instructions. But the nightly execution of the backup is a single command -- you can create a 1-line script and link a button, icon, or menu choice to it. Or try a cron entry. As for "getting in," from a remote location, with a Smoothwall and a DMZ, you can portmap to a DMZ server on the orange interface (presumes 3 NIC connections on the smoothie). Use a NONstandard SSH port on the smoothie. No sense making it obvious. Uou MUST use stricr key-based auth (which rsnapshot needs anyway). Do not allow password-based logins at all. You MUST use a forced command with the associated ssh key. Then the only thing the key does is to allow an rsync. Attempts to login are rejected. Read and understand Using Rsnapshot and SSH Now, what did I forget ? Cheers.

1. Does it boot with the new disk removed ? 2. Is there a controller BIOS setup screen (Ctrl-A at boot up, for example) ? 3. Is this a "regular" server with multiple drives, software RAID, and linux loaded ? If not, what is the manufacturer ? One thing in my experience that matches this is a system with software raid, that is trying to boot from a hard drive that was never bootstrapped. One can start up a system from drive 0. install Linux on software RAID, and think "Aha! Now I am safe!." But without loading LILO or GRUB into the boot block of the second physical hard disk, if disk 0 fails, the O/S has nothing to boot from. Another alternative, presuming SCSI disks, is the need to set the disk controller BIOS (e.g. the Adaptec 29160 or what have you) to boot from the second SCSI ID if the first one has failed and been replaced. Otherwise, the BIOS goes to the first disk, which was replaced and has no boot block. and fails to boot.

First, a caution: RAID is not a substitute for backups. It's an availability mechanism, not a disaster-recovery option. Multiple failures can and do occur at times, as our Windows administration team has found out. And more importantly, RAID does nothing to protect against inadvertant overwrite or erasure. As RAID does reduce the chance of catastrophic storage failure, it can be used to somewhat offset risk of a reduced frequency of backups, but it does not take the place of them. Put another way: With RAID 1 mirrors on my home server, I may be satified with only a weekly backup instead of a daily, since the chance of multiple disk failures is lower -- but it's still there. RAID 1 is mirroring, with or without a hotspare. The minimum is 2 disks, maximum of 3 if you can afford it. Performance is very high, since the same data is written to both disks and no calculations are necessary. But the effective total storage capacity is one half the total physical storage, or even one third if you use a hot spare. So it's more expensive. I use it (in my day job -- IT Architect) when performance is extremely important, when I only have 2 disks, or when I only have 2 or 3 I/O channels to use. Without a hotspare, when a disk fails, you stay in operation running on the remaining "mirror half," although you have lost all redundancy. Generally, you can hot-swap in a replacement drive, and the array starts chugging, reading from the good" disk and copying blocks to the new disk.. With a hotspare, this happens automatically to return the array to a state of full redundancy as quickly as possible. RAID 5 requires at least 3 disks. It breaks up the data "stripe" into 2 or more chunks (or "blocks"), and writes one block to each disk. The remaining disk gets a parity block, calculated from the data in the other blocks in the stripe. Each stripe uses the "next" disk in line for parity, so that the parity is spread over all the disks. Upside of RAID 5 is less "wasted" space -- only 1 disk's worth of space is used for parity overall, so an 8-disk set (sometimes called a "seven plus parity" or "7+P") only "wastes" 1/8th of the space. Though I'd usually set that up with a hotspare, as 6+P+HS. Downside of RAID 5 is calculation overhead. First, if you are writing a ton of data, say a transactional database, every stripe requires a new parity calculation. For most users, this is not a big deal, which is why RAID 5 is a sweet deal most of the time if you have enough resources to support it. When a disk fails in RAID 5, the system recreates the "missing" data for the stripe by calculating from the blocks on the remaining disks. this is called "degraded" mode, and you WILL notice the performance hit then. And when hotspare or replacement disk becomes available, the system begins to read EVERY block of EVERY stripe to recalculate the data to write to the new disk, called "rebuilding" the RAID set. RAID 5's can only tolerate 1 failed disk at a time. As disks get larger, the time needed to rebuild gets larger, so the risk of a second failure gets larger too. RAID 6 "wastes" more spare by storing the parity block for each stripe on two different disks. You can lose a second drive before the first one completes a rebuild, and still stay in business. We use RAID 6 for arrays built on disks larger than 750GB. BUT IT'S STILL NOT A SUBSTITUTE FOR BACKUPS. Hope this helps. Cheers.

I don't know of any slick way. With RHEL, I have to boot off the install CD or DVD and at the boot prompt do linux rescue That will boot RedHat (commercial version of CentOS), and mount the existing file systems under /mnt/sysimage That should allow you to change the root password chroot /mnt/sysimage passwd If there's an easier way, maybe someone else knows of it. But that's all I can offer. Good luck

What purpose are you formatting the drives for ? Typically, when I hear "SAS" I think "server," and that server's disk controller should do it for you. Use the server's controller setup utility (usually in the controller BIOS) to do it for you. For example, the old Adaptec controllers popped a message during boot to "Press Ctrl-A for SCSI configuration utility." Doing that brought up a menu. Our HPs have something similar, though I don't know what the current Dell PERC controllers look like or what key combo they use. If you configure them for RAID, Redundant Array of Independent Disks, see RAID - Wikipedia, the free encyclopedia and if you don't understand that after reading it, post a question and I'll try to answer it. With 5 disks, I recommend RAID 5, either 4+P or, if you can spare the space and the controller supports it, 3+P+hotspare for the best resiliency. AFTER you configure the RAID group, the RAID controller will typically present the RG as one big "disk" to the operating system. The O/S will see this as one big volume for you to use as you wish. Hope this helps

Not sure how your windows shares appear. From a command line, does 'mount' show the file system you have mounted ? If so, and if 'share1' is the only mount named like that, from a script cd `mount | grep share1 | cut -d' ' -f3` should put your script in that directory, in preparation for whatever copy or other commands you want to run. Good luck.

tar -zcvf - some_directory | ssh some_user@some_host 'cat > some_destination_file_path.tgz' Tars, compresses, and copies a directory across servers in one step. There are a number of variations on this command set, I use this one to illustrate the power of SSH since I think it's reasonably straightforward for less-experienced users to understand and adapt to their needs. Rsync can do it even better, but this forms an encrypted tunnel.

Using Clariion from old CX500s thru CX4s, and Celerra/NS NAS. If you demand performance and reliability, I consider them the ones to beat. Disclaimer: I moonlight pro bono for education, my day job is in a specialized offshoot of a large financial instution, so I have deep pockets. I have mutiple SAN farms in a number of states in the US. Never lost data due to equipment failure. My counterparts in the Windows world just lost (temporarily) 11 TB of data due to Dell MD1000 DAS array failures a month ago. Yes, they had backups, but data were inaccessible until the restores (11 TB worth in small files!) were complete. Very painful and embarassing -- and now that the horse is gone, my boss told me to move their stuff to the SAN and get them out of the storage business So reason #1 for a SAN ? Reliability. What you get for the price is redundancy, engineering, and features like scrubbing and performance analysis and improvements. 2. You get better manageability. I have over 100 servers in a typical site. The question "What's the approximate total storage in site X ?" is answered in minutes or less. That may sound trivial until you need to generate budget numbers quickly because the financial picture just changed. And maintaining 15% headroom means when my boss says "How quickly can we . . . ?" the answer is "Right away." This is also an advantage of NAS appliances. 3. You get re-use. When a need disappears, circumstances change, or I just plain goofed on the estimate, I can move storage logically thru re-zoning to meet changing conditions. This is also an advantage of NAS appliances, thru different means. 4. You get portability. If I need to upgrade a Linux server, I provision a new one, and on the appointed day I export the LUNs from one server, import them on the other, and I'm back in business quickly. I could rsync them, to be sure, but that means time spent to do it, sometimes a long time, and also the duplicate disks since the target and destination must be separate. This is also an advantage of NAS appliances. 5. Fiber channel - you get security. The FC imposes an additional security layer. Our Clariions are only accessible thru the fiber, or thru a security-enabled out-of-band network connection. Not all of these apply to schools. In fact, I doubt many schools need more than a well-designed NAS farm. A small Catholic school I know gets along very well with a open-source, linux-based NAS using several nodes. Depends on your size, your appetite, and your need for performance. Best of luck

What issue(s) are you trying to solve ? Performance woes ? Manageability ? Maintenance costs ? I generally don't recommend FC for smaller shops. Yes it's the king of security and performance, but do you really need it and do you want to pay for it ? Some parts of my shops are FC ( we're big on EMC and eminently satisfied ), with multi-TB transactional databases we need the performance. But for 4 of our satellite sites -- with 100% Windows and less that 50 servers in each -- a NAS-based solution (EMC Celeras, in fact) fills the bill nicely and leaves out the complications of additional fiber work. There's already 10G and gigabit and network expertise, so it's been mostly a plug-and-play operation. Several senior leaders "thought" we needed fiber, but I talked them out of it and now they're glad I did. For FC work, if you really want uptime, you'll want two switches, with all critical servers dual-pathed to each in case one dies, and to permit regular firmware updates. That means 2x the number of servers in switch0ports (which you will pay for), some sort of multipathing (EMC PowerPath or similar for Windows or Unix, native multipathd for Linux), and connections to each FC appliance and a dual-port FC HBA in each server. Sketch it out, add up the costs and see if you think it worth it. The NAS alternative is a second GigE NIC in each server, extra GigE blades in your network switches, and a separate VLAN for the NAS traffic to reduce contention. Do you really think you'll overwhelm that ? Do you have severe contention with your existing server-based files ? Coraid makes some interesting ATAoE stuff I've been reading about, but I have no direct experience there. I keep hoping for a new site / new project to cough up some R&D money