2 years, 9 months ago
Hello. I want to share the experience with habrovchana of expansion of a cluster of Postgres-xl in the form of the mini-instruction for "teapots". Not that there is a lot of articles and manuals on expansion of a cluster of postgres-xl, but there is enough. And in all of them is to steam of essential shortcomings according to such person as I which never was engaged before in a clustering and the more so never before worked in Linux-like axes. All articles of this sort are written for people already more or less familiar with Linux and expansion of postgresql/postgres-xl on such environment.
Therefore there was a desire will share with other practices. Further I will step by step describe all process of expansion, from downloading of source codes of postgres-xl and their compilation, before configuring of a cluster.
As many articles "for experienced" are already written, and on Habré too, I will lower the description of Postgres-xl, its components and their types (roles).
Would like to tell the history of installation of the Nominatim geocoder on dedicated server. It was initially supposed that this task will take from me about 5-7 hours, but far from it … Therefore it has been decided to write article with the description of deployment of Nominatim on the server to full operability of the site. But about everything one after another.
2 years, 10 months ago
Throughout article about the theory and practice of migration of data storages on PostgreSQL, we will talk about problems with which you can face when moving widespread MySQL DBMS. Not to tire all with excess rhetoric, today's story will be more tezisny and problem-oriented.
All below-provided is the list of standard errors of design and operation of MySQL which can influence process of adaptation of the scheme, processings of code and data transfer. Availability of all these trifles in various artful combinations is one of the reasons, on which existing? the universal? tools will hardly cope specifically with your base.
For this reason in the previous article I recommended not to spend time for search of silver bullet and to write something the? on knee? that really works. This article is urged to facilitate writing of such tool, indicating potential defects, available which you can be convinced rather quickly.
2 years, 10 months agoHi everyone! In our previous post about cloud service of Cackle we have told about architecture, technologies and loadings in general. Today we want to share how in the conditions of so strong loadings and already accumulated information (30 000 000 comments since 2011) we have made detailed analytics for system of comments of Cackle. Our method of collecting statistics universal also I think it will be interesting, in respect of practical application, all to those who has faced problem of development of analytics, but yet not absolutely represents with what to begin.
2 years, 10 months ago
Being engaged in development of the web cards using data of OpenStreetMap often there is question how to show cards with correct Russian names. This problem does not arise if your cards show only Russia. However, if you look, for example, at the map of China, you will hardly like such abundance of hieroglyphs, and vain attempts to find Beijing on such card, most likely, will not be crowned with success.
It is known that the freedom-loving OpenStreetMap project allows to save names of geographical objects in different languages. Special tags, the name:ru, name:en or name:es type are for this purpose used, and that the most important, they are filled with participants of OpenStreetMap. Of course, the most detailed texts are created by users in that language which they speak: in Russia? in Russian, in China? on Chinese, in the African countries? on local languages. Chances that some small street in Nigeria will have Russian translation, a little, but nevertheless the main geographical objects (the countries, the cities, the rivers, etc.) have transfers. This small cartographic information happens quite enough that the Russian-speaking user has opened, for example, the map of China and has found on it the main names. Thus, your resource will become a little more friendly for the user.
Documentation of receiving system of cross-compilation under Linux for Windows x32/x64/static/shared and assembly of Qt 5.4.1, last at the time of the description, in light - versions (for the specified four purposes). For itself, deep adored, well and for advantage to society.
Many developers come to conclusion that use * (in particular Linux) is more preferable nix to applications programming, using Qt framework. And to that there are reasons. Qt is initially oriented on * by nix tools, the type autotool, make, perl? And the second moment, under niksam is the fine tool? valgrind, under Windows of port did not see it so far. Well and last reason: it is simply convenient to have tools set for application creation under different target platforms? in one place.
Why the light - the Qt5 version
Freymvork-Qt has modular structure, but, alas, not the absolutely perfect. Some dependences on external libraries? are hung up? not on the module demanding these libraries, and on Qt5Core. In other words, you need this library, or not? take out yes polozh in distribution kit. Example of that? monstrik?? ICU library. It weighs nearly 25 meters! Dependence is hung up how I mentioned above, on Qt5Core? and the Qt5WebKit module demands it (which in fact in 99% of cases it is not used, at least by me). What we do? We cut and cut off. Want to receive the Qt5 full-version? It I will mention in the conclusion. Have gone.
Expensive habrachitatel, I want to share experience concerning transition, transparent for application, from PgQ queue on amqp. Perhaps it will seem by bicycle, perhaps some thoughts will be useful. Article assumes acquaintance to bases of PgQ and rabbitmq.
2 years, 11 months ago
One thing which forces me to take a detached view of ORM as they so try to hide and abstract all force and expressiveness of SQL. Before I write further, allow me to tell that, Frans Bouma has reminded me yesterday that there is difference between ORM and people who use them. It only tools (in ORM) and I agree with it, also I agree that not the bad fast food does people full, and these are people who eat it too much.
Instead of to abuse ORM I would like to tell you why I do not use OO abstraction in the basis of my database. If it is short: because SQL can strongly help to express you value of your application from the point of view of data. The only way to know as works your application this is the nobility according to what data it is generated.
2 years, 11 months ago
One of our clients has set the task — the database dump through phpPgAdmin is not filled in. In log of errors messages of type are displayed: ERROR: relation "public". "company" does not exists.
The scheme public is available, the table company is not present. In the course of search it becomes clear that via the console the same dump is filled in without problems. If to load into web form in phpPgAdmin — that too everything fulfills. But here if to load this dump into the same phpPgAdmin as the file — there are errors.
3 years ago
In recent months the perspective of migration of the working systems on solution open-source for data storage has taken minds of domestic developers. As target platform PostgreSQL enjoys special popularity. The reasons for that it is possible to call a little:
The very famous policy of import substitution staying at all implemented by the government;
Promoting of PostgreSQL by forces of enthusiasts and development of the Russian community thanks to such actions as PG Day and PGConf;
Expansion of functional capabilities of PostgreSQL allowing developers to build flexible and "schema-less" of application, without losing thus all advantages of the DBMS, such as honest transactions, fault tolerance, possibilities of scaling, etc.
We managed to be convinced of efficiency of PostgreSQL a few years ago. Implementation of the DBMS has allowed to liquidate serious technology crisis on one of large projects of the company. The detailed story about this success story has taken place on PG Day ’14 Russia, taken place last year in St. Petersburg. Since then we happened to try database for solution of wide range of problems.