Just been hectic, as I had expected. Besides the monitoring system, I've been given a couple of new tasks- one of which is to learn a new voip system. I've got to say, I'm not too terribly interested in it, as I've not been really impressed with voip. While technically it's a real time encode/decode network system, everything I've seen and experienced has led me to believe that most solutions are just barely ready. While I will admit, it's a lot easier to setup and maintain than a pbx (I still have nightmares involving analog pbx's, and horrible voicemail configurations), it just doesn't hold any real interest for me. What I'd rather be working on is the two new storage devices that we've had commissioned, plus my company's impending move to a new location. With network connections to be migrated, new services to be configured and integrated, plus new employees to train, I'm fairly excited. Unfortunately, someone's decided that voip is my new responsibility, which strikes me as a bit strange, as I've been doing firewalls, virtualization, storage and design for years. I realize I'm going to have to do this to continue being employed, but it doesn't make me want to do it any more than if I wasn't getting paid for it.
Another thing I'm hoping for is that some of my work becomes postable- right now, I can vaguely talk about it, but I can't give real specifics, or pictures. I had hoped at one point to show off some of the stuff I'm working on just because of how cool it is. Maybe in the new few months, I'll have that permission.
However, for now I've still got the usual- fires to put out, monitoring systems to maintain and build, network maintenance and storage. I want to add to that list, but so far the only thing to add is one of the things I truly don't like. Oh well, I'll do my best to try to update, but as usual... I'm swamped.
Sunday, April 22, 2012
Wednesday, April 11, 2012
Custom designing monitoring hardware
I don't know how much use this will be, but...
Current project, and probably one of the definitely more interesting ones! I get to take an atmel atmega chip, and design a monitoring system around it. Considering the price of the chip, and how easily I can get boards made (I've got a stack of custom mistakes just sitting around waiting for me to figure out what to do with them), I decided to go the route of making my own arduino compatible board (note, only compatible... it's not technically an arduino).
First task I've gotten working is monitoring temperature, power, and lock status. So far so good. Then I found the Agentuino library... which is interesting to say the least. At the moment, it only really implements two functions, namely snmp_get and snmp_set. I'll be trying to get snmp_get_next working, which would probably end up being followed by snmp_get_bulk.
Why does this even matter? Well, this system is network attached for one thing. Currently it drops it's data via a post to a LAMP server- a simple php script picks up the data and stuffs it into a mysql db. pretty simple stuff. and it does it in less than 10K of rom space. I wish I could say the same for the snmp version, as it loads out at a minimum of 14K so far. I know most people are used to working in the megs and gigs realm of storage, but I started programming on systems that had less than 16K of ram, and stored their data on tapes... I find that reducing/optimizing code to fit in the smallest of space to be an almost elegant art form. And yes, I always hear that old obi-wan quote about "An elegant weapon, for a more civilized age" whenever I think about coding for space. I would've liked to have said I would be reducing the footprint further, but with having to add roughly 30 OIDs I don't think that's going to happen.
Considering I've got about 30K of space to work with, I'm hopeful to just get that far. Adding in the fact that I'm more than likely going to have to get at least get snmp_get_next working, I may have to opt for a far bigger chip than the atmega 328. I've seen some based on the atmega2560, which has about 256K of rom, versus the 32K I'm using now. I'll have to check out the price difference as well, but I don't think it will be enough to worry about. That alone should get me a hell of a lot more space for adding in new features.
As far as why I need to get snmp_get_next working? snmpwalk and most automated NSMs require snmp_get_next in order to walk to the MIB tree. For those of you who don't know, the MIB (Management Information Base) is a specific set of data points to be returned by the device in question. Considering how many MIBs there are right now, and the fact that anyone who puts snmp support in a device tailors their own, that's a lot of MIBs in this world. And they all work off of OIDs (Object IDentifiers). The OIDs let you request specific data, and the MIB tells you how to read that data. pretty simple, right? Well, like anything out there, it's been made convoluted enough to give me a headache. There's a lot more information on the wiki about this, and described far better than I can.
Suffice it to say, I'm busy once again.
Current project, and probably one of the definitely more interesting ones! I get to take an atmel atmega chip, and design a monitoring system around it. Considering the price of the chip, and how easily I can get boards made (I've got a stack of custom mistakes just sitting around waiting for me to figure out what to do with them), I decided to go the route of making my own arduino compatible board (note, only compatible... it's not technically an arduino).
First task I've gotten working is monitoring temperature, power, and lock status. So far so good. Then I found the Agentuino library... which is interesting to say the least. At the moment, it only really implements two functions, namely snmp_get and snmp_set. I'll be trying to get snmp_get_next working, which would probably end up being followed by snmp_get_bulk.
Why does this even matter? Well, this system is network attached for one thing. Currently it drops it's data via a post to a LAMP server- a simple php script picks up the data and stuffs it into a mysql db. pretty simple stuff. and it does it in less than 10K of rom space. I wish I could say the same for the snmp version, as it loads out at a minimum of 14K so far. I know most people are used to working in the megs and gigs realm of storage, but I started programming on systems that had less than 16K of ram, and stored their data on tapes... I find that reducing/optimizing code to fit in the smallest of space to be an almost elegant art form. And yes, I always hear that old obi-wan quote about "An elegant weapon, for a more civilized age" whenever I think about coding for space. I would've liked to have said I would be reducing the footprint further, but with having to add roughly 30 OIDs I don't think that's going to happen.
Considering I've got about 30K of space to work with, I'm hopeful to just get that far. Adding in the fact that I'm more than likely going to have to get at least get snmp_get_next working, I may have to opt for a far bigger chip than the atmega 328. I've seen some based on the atmega2560, which has about 256K of rom, versus the 32K I'm using now. I'll have to check out the price difference as well, but I don't think it will be enough to worry about. That alone should get me a hell of a lot more space for adding in new features.
As far as why I need to get snmp_get_next working? snmpwalk and most automated NSMs require snmp_get_next in order to walk to the MIB tree. For those of you who don't know, the MIB (Management Information Base) is a specific set of data points to be returned by the device in question. Considering how many MIBs there are right now, and the fact that anyone who puts snmp support in a device tailors their own, that's a lot of MIBs in this world. And they all work off of OIDs (Object IDentifiers). The OIDs let you request specific data, and the MIB tells you how to read that data. pretty simple, right? Well, like anything out there, it's been made convoluted enough to give me a headache. There's a lot more information on the wiki about this, and described far better than I can.
Suffice it to say, I'm busy once again.
Saturday, April 7, 2012
SSD upgrade for my laptop plus ram...
I picked up a SanDisk Extreme 120GB SSD from Frys yesterday. Unfortunately, they didn't have what I originally wanted, which was a corsair 120GB for 100$ (99.99), so I wound up buying the next cheapest at $149... yeah, I got hosed and I know it. I also picked up 2 sticks of 4GB ram for my laptop as well, making it a pretty good upgrade round, or at least I think so anyway.
So far so good, I guess. Although it was a royal pain to get the old system migrated to the new. I made the mistake of trying to clone it using Easeus Partition Manager, which in hindsight was a massive waste of time (about 9 hours, give or take). I wound up transferring the system over using windows backup and gparted. The system is windows 7 home premium, so I ended up going through the backup control panel, and creating a system image on a usb external disk. This only took about 45 minutes and was actually faster than restoring it. The restore took roughly an hour. Once I got the restore finished, I booted off the ssd to make sure I had gotten it to work, finally... booted without issue, and pretty quickly at that... but not the "amazing, blink of an eye, instant boot" that I'd been hearing about.
Now mind you, I'd already installed the ram as well. So, less than impressed, I picked up a utility called AS SSD Benchmark, which pretty much told me where part of the problem was. I needed to align this disk. That in itself was a bit of a pain, mostly time consuming. I'd gotten a usb install of gparted live working, and used that to manipulate my partitions. Now that meant a bit of thinking, especially since nothing likes to work on the first try. I had to move the 2nd and 3rd partitions at least 20MB further from the start of the disk to make room to align the first partition, which is a bit of a weird process. To align the partition, you move the partition at least 2MB further from the start of the disk- then you move it back to 1MB away from the start, for the first partition anyway. Then the second you place 1MB away from the first, and so on until you've placed your partitions.
Depending on the size and how much data is actually there, this can take forever. Considering I had 3 partitions, but only the third really had a lot of data this was only mind numbing and not completely painful. After 3 hours, it was finally finished- remember, each partition had to be moved twice!
After that was done, rebooted into windows... well, no that didn't work. I had destroyed my boot layout during all the partition restructuring. I had to use the windows system disk I created during the system image (and yes, you'll want to create one too... if you don't, you can't restore the image anyway, so it really behooves you to do so when it asks... you'll need a blank disk for it, I used a dvd-r, but I think you can probably get away with a cd-r) to go back in and fix the boot sector and mbr map. That literally took less time to fix than it did to load the disc. Final reboot, and into windows proper again.
Much much faster this time, just the aligning made a noticeable difference at this point. However, I wasn't done yet- I needed to kill off indexing on that drive as well as automatic defragmenting. Removing the index attribute took almost 20 minutes, which again... ugh. However, at this point I decided to do some testing once it was finished. Arduino 1.0 environment took roughly 4 seconds from double click to editor ready to take code. Massive improvement already! Outlook took forever, as for some reason it had trashed it's cache, but considering that was network speed and not local disk, there wasn't anything I could do about that.
Next was firefox 11... something I have started to dread since it's been dragging terribly lately. It loaded, but no real noticeable speed increase... so I decided to load one my my bookmark groups, which was ~30 tabs. That's usually good for me to go take care of a call of nature and grab something to drink. I literally had just set the laptop down to do just that, when I saw it had loaded most the pages and only had 2 or 3 tabs left. Roughly 30 seconds to load 30 tabs... not bad. At this point, I decided to check to see if TRIM had been enabled, and nicely enough it was. Simple enough command from an elevated cmd prompt "fsutil behavior query DisableDeleteNotify", and if it returns 0, then trim is enabled for your system. Whether your device supports it or not is something you'll have to make sure of with your manufacturer. Mine as far as I can tell does.
So, keeping the old drive and the backup on standby and we'll see how long this disk lasts and whether or not it keeps me happy. For almost $200, I think it's a fairly decent upgrade. Considering how much more responsive my laptop is, I think it will be worth the price, provided the drive doesn't fall into the early death that I hear several have already.
So far so good, I guess. Although it was a royal pain to get the old system migrated to the new. I made the mistake of trying to clone it using Easeus Partition Manager, which in hindsight was a massive waste of time (about 9 hours, give or take). I wound up transferring the system over using windows backup and gparted. The system is windows 7 home premium, so I ended up going through the backup control panel, and creating a system image on a usb external disk. This only took about 45 minutes and was actually faster than restoring it. The restore took roughly an hour. Once I got the restore finished, I booted off the ssd to make sure I had gotten it to work, finally... booted without issue, and pretty quickly at that... but not the "amazing, blink of an eye, instant boot" that I'd been hearing about.
Now mind you, I'd already installed the ram as well. So, less than impressed, I picked up a utility called AS SSD Benchmark, which pretty much told me where part of the problem was. I needed to align this disk. That in itself was a bit of a pain, mostly time consuming. I'd gotten a usb install of gparted live working, and used that to manipulate my partitions. Now that meant a bit of thinking, especially since nothing likes to work on the first try. I had to move the 2nd and 3rd partitions at least 20MB further from the start of the disk to make room to align the first partition, which is a bit of a weird process. To align the partition, you move the partition at least 2MB further from the start of the disk- then you move it back to 1MB away from the start, for the first partition anyway. Then the second you place 1MB away from the first, and so on until you've placed your partitions.
Depending on the size and how much data is actually there, this can take forever. Considering I had 3 partitions, but only the third really had a lot of data this was only mind numbing and not completely painful. After 3 hours, it was finally finished- remember, each partition had to be moved twice!
After that was done, rebooted into windows... well, no that didn't work. I had destroyed my boot layout during all the partition restructuring. I had to use the windows system disk I created during the system image (and yes, you'll want to create one too... if you don't, you can't restore the image anyway, so it really behooves you to do so when it asks... you'll need a blank disk for it, I used a dvd-r, but I think you can probably get away with a cd-r) to go back in and fix the boot sector and mbr map. That literally took less time to fix than it did to load the disc. Final reboot, and into windows proper again.
Much much faster this time, just the aligning made a noticeable difference at this point. However, I wasn't done yet- I needed to kill off indexing on that drive as well as automatic defragmenting. Removing the index attribute took almost 20 minutes, which again... ugh. However, at this point I decided to do some testing once it was finished. Arduino 1.0 environment took roughly 4 seconds from double click to editor ready to take code. Massive improvement already! Outlook took forever, as for some reason it had trashed it's cache, but considering that was network speed and not local disk, there wasn't anything I could do about that.
Next was firefox 11... something I have started to dread since it's been dragging terribly lately. It loaded, but no real noticeable speed increase... so I decided to load one my my bookmark groups, which was ~30 tabs. That's usually good for me to go take care of a call of nature and grab something to drink. I literally had just set the laptop down to do just that, when I saw it had loaded most the pages and only had 2 or 3 tabs left. Roughly 30 seconds to load 30 tabs... not bad. At this point, I decided to check to see if TRIM had been enabled, and nicely enough it was. Simple enough command from an elevated cmd prompt "fsutil behavior query DisableDeleteNotify", and if it returns 0, then trim is enabled for your system. Whether your device supports it or not is something you'll have to make sure of with your manufacturer. Mine as far as I can tell does.
So, keeping the old drive and the backup on standby and we'll see how long this disk lasts and whether or not it keeps me happy. For almost $200, I think it's a fairly decent upgrade. Considering how much more responsive my laptop is, I think it will be worth the price, provided the drive doesn't fall into the early death that I hear several have already.
Thursday, March 29, 2012
And tonight it's small scale VCenter DB maintenance.
A customer of mine wanted to use VMWare VCenter 4.1 a while back, so I installed and configured it for them following their specifications. Of course, I monitor the living daylights out of it too. Tonight vpxd started failing repeatedly. A quick check into this shows that the DB is full. The error message in question is quite literally "CREATE DATABASE or ALTER DATABASE failed because the resulting cumulative database size would exceed your licensed limit of 4096 MB per database."
Now given that it's using SQL express instead of a full SQL install, that's pretty understandable. However, my joy stemmed from the fact that I couldn't run the stored procedures to clean it up (DB was literally too full), so I wound up having to clean the DB using a script from VMWare's help site, specifically the one found here. If you haven't already done so, you'll need to install Microsoft SQL Server Management Studio Express. It's fairly simple and straight forward. Once you're finished, you'll need to run the sql script you downloaded earlier. Make sure you are logged in with the right permissions, otherwise the script will fail repeatedly. Also, make sure you select the correct DB before running the script.
There is one gotcha with the script, you do need the change at least one parameter, possibly more depending on how much data you want to clear.
-- ######### USER CONFIGURABLE PARAMETERS ########################
-- 0 = COUNT ONLY; 1 = DELETE ROWS
SET @DELETE_DATA = 0
-- Use one of these methods to specifiy the data cutoff date
SET @CUTOFF_DATE = GETUTCDATE()-180
--SET @CUTOFF_DATE = '2007/01/01'
-- Number of rows to delete per transaction
SET @BATCH_SIZE = 10000
-- ######### END USER CONFIGURABLE PARAMETERS ####################
Make sure you change @DELETE_DATA to 1, otherwise this script won't do anything. I also had to change the cutoff date calculation to 30 due to the fact that I had a lot of activity in the last 3 months. And I do mean a lot- over 6 million entries.
I shouldn't have been surprised by this, but it came nearly 6 months sooner than I expected it to. Good luck with your own maintenance!
Now given that it's using SQL express instead of a full SQL install, that's pretty understandable. However, my joy stemmed from the fact that I couldn't run the stored procedures to clean it up (DB was literally too full), so I wound up having to clean the DB using a script from VMWare's help site, specifically the one found here. If you haven't already done so, you'll need to install Microsoft SQL Server Management Studio Express. It's fairly simple and straight forward. Once you're finished, you'll need to run the sql script you downloaded earlier. Make sure you are logged in with the right permissions, otherwise the script will fail repeatedly. Also, make sure you select the correct DB before running the script.
There is one gotcha with the script, you do need the change at least one parameter, possibly more depending on how much data you want to clear.
-- ######### USER CONFIGURABLE PARAMETERS ########################
-- 0 = COUNT ONLY; 1 = DELETE ROWS
SET @DELETE_DATA = 0
-- Use one of these methods to specifiy the data cutoff date
SET @CUTOFF_DATE = GETUTCDATE()-180
--SET @CUTOFF_DATE = '2007/01/01'
-- Number of rows to delete per transaction
SET @BATCH_SIZE = 10000
-- ######### END USER CONFIGURABLE PARAMETERS ####################
Make sure you change @DELETE_DATA to 1, otherwise this script won't do anything. I also had to change the cutoff date calculation to 30 due to the fact that I had a lot of activity in the last 3 months. And I do mean a lot- over 6 million entries.
I shouldn't have been surprised by this, but it came nearly 6 months sooner than I expected it to. Good luck with your own maintenance!
Tuesday, March 27, 2012
And once again, I've managed to wind up too busy...
I will try to keep this updated far better than I have in the past, however I've just gotten a ton of new responsibilities, so the chances of update is going down severely.
Thursday, March 22, 2012
Three new spiceworks related projects in the near future
Which will probably culminate in the creation of a new blog while I teach myself how to do this (and share whatever I manage to learn with everyone else)-
Project 1: create plugin that will pull data from custom monitoring database, and allow it to be displayed in spiceworks, either as a dashboard widget, or as custom attributes on each related items entry. preferably both, with the creation of a rack level object as well.
Project 2: create plugin that will handle ticket expiration and escalation. By putting a due date on the ticket, the plugin will scan every 15 minutes, checking each open ticket for expiration. at a configurable time remaining point, an email will be sent out to the ticket assignee reminding them to get back to the ticket. If no action is taken by the time the ticket expires, the ticket will be reassigned to the next higher level.
Project 3: create plugin that will manage ticket workflow. This still hasn't gotten a good definition yet, so it's in limbo. problem is, this is the one that's due first. Amazing how that works isn't it?
If you haven't used, seen or heard of Spiceworks yet, go check it out! it's a great piece of software, and can make managing your network much less of a headache. If you're already using other solutions, I'd still check it out anyway, as there's always the chance you might find something you like better, or a feature you want that doesn't exist in yours. Or vice versa, there's nothing wrong with competition!
Project 1: create plugin that will pull data from custom monitoring database, and allow it to be displayed in spiceworks, either as a dashboard widget, or as custom attributes on each related items entry. preferably both, with the creation of a rack level object as well.
Project 2: create plugin that will handle ticket expiration and escalation. By putting a due date on the ticket, the plugin will scan every 15 minutes, checking each open ticket for expiration. at a configurable time remaining point, an email will be sent out to the ticket assignee reminding them to get back to the ticket. If no action is taken by the time the ticket expires, the ticket will be reassigned to the next higher level.
Project 3: create plugin that will manage ticket workflow. This still hasn't gotten a good definition yet, so it's in limbo. problem is, this is the one that's due first. Amazing how that works isn't it?
If you haven't used, seen or heard of Spiceworks yet, go check it out! it's a great piece of software, and can make managing your network much less of a headache. If you're already using other solutions, I'd still check it out anyway, as there's always the chance you might find something you like better, or a feature you want that doesn't exist in yours. Or vice versa, there's nothing wrong with competition!
Neat little trick I learned tonight
And it seems fairly stupid too-
For some reason, I couldn't replace the owner on a file using powershell directly. Doubting myself, I went and found a script someone else had written (powershell team blog), and that failed as well! The error given is "Set-Acl : The security identifier is not allowed to be the owner of this object.". Well that sucks. So I did some more digging, and came across a solution- apparently windows server 2008 will let you do this, if you use the UNC path for your filenames. Why that makes a difference, who knows.
Any why was any of that important? because like a dummy, I blew away a hyper-v virtual machine without realizing I wasn't done with it, and had to recover it using Piriform Recurva. Great product, but it didn't restore the ACL's correctly, and I couldn't reassign the ACL directly. So I had to copy them from a file that still had the correct ACL. Worked like a charm, and now I'm happy once again. Well, as happy as I get anyway.
Where did I get the idea? from Fixing IT. Thanks, I appreciate it.
For some reason, I couldn't replace the owner on a file using powershell directly. Doubting myself, I went and found a script someone else had written (powershell team blog), and that failed as well! The error given is "Set-Acl : The security identifier is not allowed to be the owner of this object.". Well that sucks. So I did some more digging, and came across a solution- apparently windows server 2008 will let you do this, if you use the UNC path for your filenames. Why that makes a difference, who knows.
Any why was any of that important? because like a dummy, I blew away a hyper-v virtual machine without realizing I wasn't done with it, and had to recover it using Piriform Recurva. Great product, but it didn't restore the ACL's correctly, and I couldn't reassign the ACL directly. So I had to copy them from a file that still had the correct ACL. Worked like a charm, and now I'm happy once again. Well, as happy as I get anyway.
Where did I get the idea? from Fixing IT. Thanks, I appreciate it.
Subscribe to:
Posts (Atom)