Which OS is better to use to create a cluster

There is a question about choosing an OS to create a cluster using ISPmanager Cluster. If anyone had or has experience in this area, I will be happy to listen and take into account. Thank. Read more

Replication Redis

I am engaged in one startup in which we use radishes, now the question of building a fault-tolerant database cluster has come close. As is known, radish currently supports only master-slave replication. It is necessary that if a master falls, one of the slaves would take the function of a master. For this purpose, I found the following solution: When the master crashes, this monitor selects a new master and sends a synchronization signal with it to t... Read more

Fail Over Cluster on Windows Server 2008r2 and Oracle Fail Safe Manager

The task was to install the project on Fail Safe Cluster on the following tools: Windows Server 2008 (already configured Fail Over two-node cluster) and Oracle 11g2 Standard Edition One. As far as I understand, the Fail Over OS level should deal with the start and stop of the necessary services. Why then does Oracle provide a special utility Fail Safe Manager? Do I need to install a utility for the database? Strictly speaking, I have already installed the utility on the nodes, but did no... Read more

Marker manager (clusterer) for the Yandex.Maps API?

Hello everyone, For Google Maps second version, I used such a clusterer that automatically hid or showed markers on the map, thus ensuring performance: Tell me, is there a similar clusterer (marker manager) for Yandex.Maps? It is desirable that this was a ready-made solution. Thank you in advance. Read more

DRBD dual primary + iscsi multipath?

Question: if a resource is synchronous (protocol Z) is replicated between two servers, each of which has an independent iscsi target for this resource, can these two targets be exported as a single resource using mutipath? Reading Lkml has so far suggested that the synchronization mode is important. What other moments might be? Read more

Who generally needs clusters?

Good day. Appeared the ability to collect linux cluster. Small, 64 cores of 5300 bahomips, 48 ​​gigabytes of memory, about 30 terabytes. You can collect something. Who to offer? Do you know anyone who needs clusters for white technologies (that is, not for bridging md5)? Read more

Calculation of the cost of a computing cluster?

Interested in the issue of scalability and technical organization of cluster calculations. Namely, how does the cost of creating a cluster change depending on the number of computers in it? For example, I can assemble a small cluster of 3-5 computers at home and its cost will be equal to the price of computers. If you increase the number of computers at home to build such a cluster does not work, you will need to rent a room with enough power supply, racks, cooling systems, fire suppressi... Read more

how to generate a unique integer id in a cluster?

Hello, I would like to understand the issue of generating a unique numeric ID in the cluster. Actually looking at while the same option is seen, only the size of the allocated bytes will be slightly less. We do 4 parts: 1) time in epoch format (unix, posix), in total the number of seconds since 1970. time = 8 bytes integer (2 ** (4 * 8) = 4294967296 variants) import time int (time.time ()) part1 = 1314823196 / ... 2) machine_id ... Read more

windows 7 instead of windows xp?

On the computer, the HDD is divided into two sections: the logical drive C is 14.6 GB (Windows XP is on it and the programs) and the logical disk D is 134 GB (other files are on it). NTFS file system, cluster size 4.00 KB. It is planned to install Windows 7. The ability to backup the necessary files is. How to do it better? Namely: Option 1: Re-divide the disk into sections? If yes, then two or three sections already (1st - for Windows 7, 2nd - for programs, 3rd - for other files)? If two sect... Read more

Django + Celery + RabbitMQ in a cluster. In the queue we have duplicates from the tasks periodic_task

Two copies of Django (1.3) applications are running on two nodes. Celery (2.4) with backend RabbitMQ (2.7.1) are installed on two. RabbitMQ replicated with each other. The application has periodic tasks. Let's say every hour you need to download and parse the document. So at the same time, each of the two applications adds to the queue according to the task (according to the same task). And so we download the document twice and parse it twice.   How to avoid it ??? Read more

Useful Links