2014/10/23
by Piotr Pisz
0 comments
Much of the servers used by us in the Datacenter is equipped with internal hard drives. They are not used for anything. I decided to take the opportunity to move these drives to our Lab in order to better test the EMC ScaleIO. Before physically removing, done on each server detach procedure. In some cases, this procedure ended with the message “The device does not contain a diagnostic partition”:
2014/10/21
by Piotr Pisz
24 Comments
Often we do not have access to the specific vendor hardware. Storage are very expensive toys and do not appear, just like that, in home labs. Fortunately, companies such as EMC provide simulators. VNX simulator, for example. Of course, this product is devoid of block parts (FC) but it is a fully functional file-part (including the full management interface). Completely free. What does that give to us? Ability to check in practice as it looks professional disk array and the possibility to test several solutions. Replication between arrays (this requires a two simulators) for example. And of course, satisfy curiosity as always.
2014/10/10
by Piotr Pisz
0 comments
This week started for us terribly, on Monday random View machines has started restart cycle, interrupted by “Blue Screen of Death”. The first suspicion fell on the Trend Micro Deep Security 9.5 and its new modular agent. The driver what caused the error “DRIVER_IRQL_NOT_LESS_OR_EQUAL” is tbimdsa.sys (one of the main drivers from Deep Security 9.5). The problem was very difficult to diagnose, the machine from time to time run in a safe mode but mostly run with BSOD. Internet search slowly indicating that the problem may also be in a server. Experimentally, one of the machines that refused to run, has been moved to the HP server. And launched without a problem!
Now it turned out that the error is in the BIOS on Dell R720 servers and concerns memory access (random machines reboots and other errors like IRQL). If you do not have loaded latest firmware (2.2.3) do it as soon as possible:
PowerEdge R720/R720-xd BIOS Version 2.2.3
Updated Intel Xeon processor E5-2600 V2 product family microcode to 427.
Fixes
=====
Corrected an issue where a complex sequence of internal processor events may result in unexpected page faults or use of incorrect page
Translations If EPT (Extended Page Tables) is enabled . Due to this erratum a guest may crash or experience unpredictable system behavior.
After uploading a new BIOS all the problems with the machines ran out.
2014/09/22
by Piotr Pisz
2 Comments
Before writing this post I made a little research. I came out to me that it is not so easy to arrange remote access through the web browser to your own workstation. There are various solutions, for the most demanding installation of the server and the client. The closest solution to this which I will describe here, is probably “Microsoft Remote Desktop Web Access”, but it is a feature of Microsoft Server. And now we have a premiered VMware Horizon 6 with View, and with it a novelty, Horizon View HTML Access. Under this name hides the BLAST protocol, which until now has been part of Horizon View Connection Server and which allow holders of View infrastructure comfortable working on workstations from anywhere via a web browser. Today has been separated out (View construction has become more modular), installation and configuration is not simple (but also without exaggeration), but its completion will allow us, for free, enjoy remote access to the desktop. I write it for free, of course, you first need to download Horizon 6 with View the 60 day license, but Direct Connect itself does not require installed and valid license. Besides, VMware has a very liberal attitude to the “Home Lab” and does not make the problems people gaining experience in the field of their products.
2014/09/12
by Piotr Pisz
0 comments
With so-ranking title, must immediately answer the question of why we have to balance traffic without so cool solution which is the iApp (for View)? The answer is simple, using iApp, do not do anything you did not anticipate its creator (for example, did not anticipate the use of security policies, and there is no BLAST protocol handler). The main advantage of iApp is the speed of implementation, a few clicks and you’re done. Manually takes a little longer but the effect is the same and the possibility is much larger. In the configuration I have described, I used the balancing of traffic at connection server level without security servers. I believe that balancing security servers does not make sense, in the end F5 has to replace them! Of course in our network are such servers, they are the traditional security gateway for privileged users (VIP), which is mainly Administrators. The same applies to the internal network, can simultaneously connect through F5 (User) or directly to the connection server (VIP). This configuration gives us a high degree of flexibility and resilience to potential failures and a lot of opportunities for Administrators.
2014/09/07
by Piotr Pisz
0 comments
In the first part I described how to configure Nested ESXi, and additional entry described how to configure the network, so that in vCenter 5.5 was possible to communicate with it. So I assume that we have our wonderful, active, virtual laboratory. This is a really great solution and I highly recommend to playing with it. However, after prolonged use in vSphere in which they operate, we can observe with surprise that the network traffic directed to the Nested ESXi is very large. It is associated with the enabled “promiscuous mode”, in this mode, the network card address all network traffic. In my company, in physical lab I have only one nested environment with four hosts. From time to time I do on them various tests. Yet, as I look at the network load charts, that generates vCenter Operations Manager, my Nested ESXi are at the forefront! In this figure it is clearly visible (PESX1-4 is a Nested ESXi):
