Boogie-woogie-hoster or walking over three seas-3

    The story of finding the best hoster.
    Moscow-Erfurt-Moscow.

    Part one. Moscow


    Less than a year ago, I needed a hoster for my project. By the will of fate, at that time my friend worked in the support of the HC. Without thinking twice, I settled on a shared from the HC. The project was small, it worked fine and I calmed down for a while. After some time, the limit on the number of file accesses began to be exceeded, and I had to switch to VPS from the same hoster.

    Switching to VPS was very difficult. Then I still did not know Linux very well, and installing the Apache + Nginx + Memcached bundle was somewhat problematic. Then it was required to establish statistics of Awstats. Then fasten a couple more sites to the existing hosting. A separate problem was the installation of mail. There are many domains, there were more than one mailbox per domain. Organizing all this on an existing base without certain knowledge of the hands through the console was quite difficult. And finally, everything is installed, screwed, everything works.

    But after a while, the server began to crash. Periodically. With a frequency of once a day. The number of letters with the subject “Server is not working” began to exceed 10 per month, and a friend from support was fired four months ago. No matter what I try to do, no matter how I monitor it, the feeling is that some cache clogged all the memory or gc worked poorly. Support wrote that every day (with a load of 100-150 unique per day) the limit in user_beancounters was exceeded. Sometimes the picture was this: After a while I started getting MySQL errors: An error occurred while executing a complex instruction with many JOIN, HAVING and WHERE. No matter how I tried to repeat it on Denver - to no avail. But I don’t think that sampling from ~ 800 rows with LIMIT 0, 4 can kill MySQL so cool.
    resource held maxheld barrier limit failcnt
    kmemsize 4719688 8389126 8388608 8388608 24901153
    numtcpsock 12 120 120 120 3628
    tcpsndbuf 3416 482996 478932 786432 108970
    tcprcvbuf 0 514420 478932 786432 645



    Error Number: 1030
    Got error -1 from storage engine




    After a long optimization of MySQL, nginx and apache it became a little better - errors climbed in a day. Apache could not fork new processes due to lack of memory, nginx went into loop. Remembering the words of my friend, “All developing projects eventually go from HC to the West,” I decided - it’s time!

    Part two. Erfurt


    Many experts in the hosting world could already guess where I landed, but we will not get ahead of ourselves. Having once again studied the situation around the world of hosters, I settled on the most (as it seemed to me then) simple and convenient, and, most importantly, cheap, option - KEYWEB. I just needed to - Russian or English support, linux vps, root-SSH, the ability to add ~ 10 sites, the same database, three times more mailboxes, backups and virtualization technology not OpenVZ. After registering, I was offered to pay the bill within 7 days, and provided me with an account to transfer my project to their site.

    Plesk. Those who have previously worked with their hands through 22 holes will understand me - this is relief, a sense of comfort and convenience. Everything is done through a beautiful interface. To check the transition of the zone, BIND is installed, it is possible to install nginx with one button, separation of the database and mail into domains, different FTP access for each domain, well ... who worked, knows. Another cloud of features and lotions. I transferred all my sites, changed the DNS, and by going to the technical address, I made sure that it worked. There were, of course, problems with support, such as, for example, the answer is not my question, but are these the little things?

    I liked everything, and I went to pay the bill. Keyweb`s invoices are sent to Germany, therefore, instead of the usual BIC, R / C and other things, they have BiC, BLZ and other abbreviations incomprehensible to the average citizen. In my great naivety, I assumed that Sberbank had professional specialists who could transfer money abroad, all the more so since they always have the stickers “transfer abroad”, “transfer abroad”. But I could not imagine that the entire Sberbank team would look at the bill that was billed to me, surprisingly scratching their turnips. I was advised to contact another bank.

    Since Sberbank is in the immediate vicinity of my house, and any other bank (Legion or VTB24) can be reached either in 15 minutes on foot or in a couple of minutes by car, I decided to choose the latter and ... didn’t get there! Only the first snows had left, but in some places there was still ice porridge, water and ice, as when leaving the garage, the car sat on the bottom. And neither there nor here. Help has arrived. Of course, we pulled out the car, but "the sediment remained." It was a sign. As a result, I did not pay the bill - the zone switched to a German server.

    And here he is - a terrible moment of awakening from the happiness and joy of Plesk. 6 second ping. Those. at least six second ping. The minimum page opening time was 6 seconds.It happened that the page opened in 10 seconds. And that damn did not suit me. I decided to wait another day and made sure of this wonderful phenomenon of servers in the city of Erfurt - 6 seconds.

    Part three. Moscow


    Returning a bad omen. In addition, in the HC. It was decided to choose something else, since such experiments had begun. A friend who left the HC advised a new hoster - Logol. These are the guys from the HC who decided to create the best hosting provider and founded Logol. Walking around their site, asking the online consultant, I was hooked by user_beancounters, which they suggested. Somewhere they were 2 times, more than in the HC, somewhere in 3-4.

    Almost before pressing the “order!” Button, something skipped a beat somewhere, and I turned my attention to another hoster - Agave. She has twice as many Beancounters as Logol, however, it costs a little more. But there it’s tricky: either Debian and ISPManager Lite (+ 100rub), or CentOS and pens (their own - for free). The RAM, however, is 2 times less than that of the HC, but ... having decided that I should try to the end, I took hosting in Agave.
    ISPManager Lite pleased with its simplicity. Backups were done right in, the installation of the necessary components and mailboxes for individual domains were created at the click of a button. Comfort and happiness. Not as cool as in Plesk, but no more.

    True, there were uncomfortable and frankly terrible moments in Agave's personal account. For example, domain join killed: you need to choose the hosting to which you attach the domain, select the domain (in the .ru zone - 150 rubles, in .com - 600 rubles), select the payment method (sber, money, webmoney, account in another bank ), click to create a new order and only then see that the cost of services is 0 rubles.

    But now my sites just fly. Fractions of a second - and voila! I even forgot the Linux commands!

    Conclusion. Hosters are different


    ХЦ has lost all my trust, constant unsubscribe support, that I am a fool, constant problems with the server at 150 people per day, constant server reboot, scripts that work everywhere, in the ХЦ give a MySQL error.

    Foreign countries are not good for everyone either. The plus is that we are completely stable ... guardians will not come to arrange mask shows, for example, in the city of Erfurt. But problems with payment and high pings, high page opening times make it difficult to rejoice.

    I am not talking specifically about Agave, on Logole I also have a small site, and it works well. I used to choose the Majordomo as the most adequate. Many sites that I made are hosted there, because there "there were no gaps at all." After all, I, as a web developer, advise my clients and hosting providers as well.

    This is my story.

    Also popular now: