April 29, 2019 at 11:471C Developer Tales: admin
All 1C developers in one way or another closely interact with IT services and directly with system administrators. But not always this interaction goes smoothly. I would like to tell you some funny stories about this.
Most of our customers are large holdings with their large IT departments. And for backup copies of infobases, as a rule, the client’s specialists are responsible. But there are relatively small organizations. Especially for them, we have a service according to which we take care of all issues related to backing up all 1C. About such a company will be discussed in this story.
A new client came in support of 1C, and, among other things, the contract had a clause that we were responsible for the backups, although they had their own system administrator on staff. The client-server database, as the DBMS - MS SQL. A fairly standard situation, but still there was one caveat: the main base was quite large, but at the same time the monthly increase was very small. That is, the database contained a lot of historical data. Given this peculiarity, I set up backup maintenance plans as follows: on the first Saturday of each month a full backup was made, it was quite weighty, then a differential copy was made every night - a relatively small amount and every hour a copy of the transaction log. Moreover, full and differential copies, not only that were copied to a network resource, it was also additionally uploaded to our FTP server. This is a mandatory requirement for the provision of this service.
All of this was successfully configured, put into operation and worked in general without fail.
But a few months later, the system administrator changed in this organization. The new system administrator began to gradually rebuild the company's IT infrastructure in accordance with modern trends. In particular, virtualization appeared, disk shelves, access was closed everywhere and everything, etc., which in the general case, of course, cannot but rejoice. But it did not always go smoothly with him, often there were problems with the performance of 1C, which caused some disagreement and misunderstanding with our support. Also, it should be noted that our relations with him generally developed quite cold and somewhat strained, which only increased the degree of tension in case of any problems.
But one morning it turned out that the server of this client was unavailable. I called the system administrator to find out what happened and received something like “Our server has crashed, we are working on it, it’s not up to you.” Well, that works. So the situation is under control. After lunch, I call back again, in the administrator’s voice, instead of irritation, I already feel tiredness and apathy. Trying to clarify what happened, and can we somehow help? The conversation revealed the following:
He moved the server to a new storage system with a raid just assembled. But something went wrong and after a few days this raid crumbled safely. Either the controller burned out, or something happened to the disks, I don’t remember exactly, but all the information was irretrievably lost. And the main thing was that the network resource with backups also in the process of any migration was on the same disk array. That is, both the productive base itself and all its backups were lost. And what to do now is not clear.
Calmly, I say. We have your night backup. The answer is silence, by which I understand that I just saved a person’s life. We begin to discuss how to transfer this copy to a new, just deployed server. But here a problem arose.
Remember, I said that the full backup was quite large? I did it for a reason once a month on Saturdays. The fact is that the company was a small plant, which was located far beyond the city and the Internet they had was very so-so. By Monday morning, that is, over the weekend, this copy with grief in half had time to upload to our FTP server. But there was no way to wait a day or two until it loads in the opposite direction. After several unsuccessful attempts to drag the file, the administrator took out the direct hard drive from the new server, found a car with a driver somewhere and quickly rushed to our office, since we are still in the same city.
While we were standing in the server room and waiting for the files to be copied, we first met, so to speak, “live”, drank a cup of coffee, talked in an informal setting. I sympathized with his grief and sent back with a full screw of backups, in a hurry to restore the stopped work of the company.
Subsequently, all our applications to the IT department were resolved very quickly and there were no more disagreements.
Once, with a single client, for a very long time I could not publish 1C for web access through IIS. It seems to be an ordinary task, but here it didn’t go out of the way. Local system administrators connected, tried different settings and configuration files. 1C on the web normally did not want to work in any. Something was wrong with domain security policies, or with a local sophisticated firewall, or what the hell. At the N-th iteration, the admin drops a link to me with the words:
- Try again using this instruction. Everything is pretty detailed there. If it doesn’t work out, write to the author of this site, maybe he will help.
“No,” I say, “it will not help.”
- Why?
- I am the author of this site ... (
As a result, they launched Apache without any problems. IIS could not win.
We had a client - a small manufacturing enterprise. They had a server, such a peculiar “classic” 3 in 1: terminal server + application server + database server. They worked in some industry configuration based on soft starters, there were about 15-20 users in the system, and the system performance was in principle suitable for everyone.
Time passed, everything worked more or less stable. But Europe imposed sanctions against Russia, as a result of which the Russians began to buy mainly domestic products, and things at this company went uphill. The number of users has grown to 50-60 people, a new branch has opened, and so has the workflow. And now the current server has ceased to cope with the sharply increased load, and 1C started, as they say, "slow down". At peak hours, documents were held for several minutes, blocking errors rained down, forms opened for a long time, well, and the whole other bunch of related services. The local system administrator dismissed all the problems, saying, "This is your 1C, you have to figure it out." We have repeatedly suggested conducting an audit of the system for performance, but this didn’t reach the audit itself.
Well, I sat down and wrote a rather voluminous letter stating that it is necessary to separate the roles of the terminal server and the application server from the DBMS (which, in principle, we have already said repeatedly before). I wrote about DFSS on terminal servers, about Shared Memory, threw links to authoritative sources and even offered some hardware options. This letter reached the powers that be in the company, went back to the IT department with the “Execute” resolutions, and the ice generally broke.
After some time, the admin sends me the IP address of the new server and login credentials. He says that MS SQL and the components of the 1C server are deployed there, and you need to transfer the databases, but so far only to the DBMS server, as there were some problems with the 1C keys.
Indeed, all the services came in, the server is not very powerful, okay, I think it's better than nothing. I’ll transfer the databases so far to at least somehow relieve the current server. In the negotiated time, he performed all the transfers, but the situation did not change - all the same performance problems. Strange, of course, well, let's register the bases in the 1C cluster, we'll see.
It takes several days, the keys have not been transferred. I’m interested in what the problem is, everything seems to be simple there - I removed it from one server, stuck it in another, installed the driver and it’s ready. The admin responds, says something about port forwarding, a virtual server, and more.
Hmm ... a virtual server? It seems that there has never been any virtualization and they didn’t exist ... I recall a rather well-known problem with the inability to forward the 1C server key to the virtual machine on Hyper-V in Windows Server 2008. And here some suspicions begin to build up ... I
open the server manager - Roles - a new one role is Hyper-V. I go to the Hyper-V manager, I see one virtual machine, I’m connecting ... And really ... Our new database server ...
So what? The instructions of the authorities and my recommendations are fulfilled, the roles are separated. The task can be closed.
After some time, the crisis now happened, the new branch had to be closed, the load decreased, the system performance became more or less tolerable.
Well, of course, they could not forward the server key to the virtual machine. As a result, everything was as it was and left: the terminal server + 1C cluster on the physical machine, the database server in the same place in the virtual one.
And okay, would it be some sort of Sharashkin office. So no. A well-known company, the products of which you probably know and have seen in the corresponding departments of all kinds of Ribbons and Auchanov.
One large holding company with ambitious plansto take over the world once again bought a small company with the goal of incorporating it into its megacorporation. In all divisions of this holding, users work in their databases, but with an identical configuration. And so we started a small project to include a new unit in this system.
First of all, it is necessary to deploy combat and test bases. The developer received the data for the connection, logs on to the server, sees the installed MS SQL, server 1C, sees 2 logical drives: a 250 gigabyte C drive and a 1 terabyte D drive. Well, “C” is a system, “D” for data, the developer logically decides and deploys all the databases there. I even set up maintenance plans, including backups, just in case (although we are not responsible for this). True backups took shape here on the "D". In the future, it was planned to reconfigure already on some separate network resource.
The project was launched, consultants conducted training on how to work in the new system, leftovers were transferred, some small point improvements were made and users began to work already in the new information base.
Everything went well until one morning on Monday it was discovered that the database drive had disappeared. There is simply no “D” on the server and that’s it.
Further investigation revealed this: in fact, this “server” was the working computer of the local system administrator. True, the server OS was still on it. A personal USB disk of this admin was inserted into the server. And the administrator went on vacation taking with him his own screw, with the aim of pumping films on it on the road.
Thank God, he did not manage to delete the database files and managed to restore the working database.
It is noteworthy that everyone, in general, was satisfied with the performance of the system located on the USB-drive. Nobody complained about any unsatisfactory work of 1C. It was already later that the megaproject began at the holding to transfer all information databases to a single centralized platform with super-servers, storage for a million + rubles, sophisticated hypervisors and unbearable 1C brakes in all branches.
But this is a completely different story ...
High-speed communication channel
Most of our customers are large holdings with their large IT departments. And for backup copies of infobases, as a rule, the client’s specialists are responsible. But there are relatively small organizations. Especially for them, we have a service according to which we take care of all issues related to backing up all 1C. About such a company will be discussed in this story.
A new client came in support of 1C, and, among other things, the contract had a clause that we were responsible for the backups, although they had their own system administrator on staff. The client-server database, as the DBMS - MS SQL. A fairly standard situation, but still there was one caveat: the main base was quite large, but at the same time the monthly increase was very small. That is, the database contained a lot of historical data. Given this peculiarity, I set up backup maintenance plans as follows: on the first Saturday of each month a full backup was made, it was quite weighty, then a differential copy was made every night - a relatively small amount and every hour a copy of the transaction log. Moreover, full and differential copies, not only that were copied to a network resource, it was also additionally uploaded to our FTP server. This is a mandatory requirement for the provision of this service.
All of this was successfully configured, put into operation and worked in general without fail.
But a few months later, the system administrator changed in this organization. The new system administrator began to gradually rebuild the company's IT infrastructure in accordance with modern trends. In particular, virtualization appeared, disk shelves, access was closed everywhere and everything, etc., which in the general case, of course, cannot but rejoice. But it did not always go smoothly with him, often there were problems with the performance of 1C, which caused some disagreement and misunderstanding with our support. Also, it should be noted that our relations with him generally developed quite cold and somewhat strained, which only increased the degree of tension in case of any problems.
But one morning it turned out that the server of this client was unavailable. I called the system administrator to find out what happened and received something like “Our server has crashed, we are working on it, it’s not up to you.” Well, that works. So the situation is under control. After lunch, I call back again, in the administrator’s voice, instead of irritation, I already feel tiredness and apathy. Trying to clarify what happened, and can we somehow help? The conversation revealed the following:
He moved the server to a new storage system with a raid just assembled. But something went wrong and after a few days this raid crumbled safely. Either the controller burned out, or something happened to the disks, I don’t remember exactly, but all the information was irretrievably lost. And the main thing was that the network resource with backups also in the process of any migration was on the same disk array. That is, both the productive base itself and all its backups were lost. And what to do now is not clear.
Calmly, I say. We have your night backup. The answer is silence, by which I understand that I just saved a person’s life. We begin to discuss how to transfer this copy to a new, just deployed server. But here a problem arose.
Remember, I said that the full backup was quite large? I did it for a reason once a month on Saturdays. The fact is that the company was a small plant, which was located far beyond the city and the Internet they had was very so-so. By Monday morning, that is, over the weekend, this copy with grief in half had time to upload to our FTP server. But there was no way to wait a day or two until it loads in the opposite direction. After several unsuccessful attempts to drag the file, the administrator took out the direct hard drive from the new server, found a car with a driver somewhere and quickly rushed to our office, since we are still in the same city.
While we were standing in the server room and waiting for the files to be copied, we first met, so to speak, “live”, drank a cup of coffee, talked in an informal setting. I sympathized with his grief and sent back with a full screw of backups, in a hurry to restore the stopped work of the company.
Subsequently, all our applications to the IT department were resolved very quickly and there were no more disagreements.
Contact your system administrator
Once, with a single client, for a very long time I could not publish 1C for web access through IIS. It seems to be an ordinary task, but here it didn’t go out of the way. Local system administrators connected, tried different settings and configuration files. 1C on the web normally did not want to work in any. Something was wrong with domain security policies, or with a local sophisticated firewall, or what the hell. At the N-th iteration, the admin drops a link to me with the words:
- Try again using this instruction. Everything is pretty detailed there. If it doesn’t work out, write to the author of this site, maybe he will help.
“No,” I say, “it will not help.”
- Why?
- I am the author of this site ... (
As a result, they launched Apache without any problems. IIS could not win.
Deeper
We had a client - a small manufacturing enterprise. They had a server, such a peculiar “classic” 3 in 1: terminal server + application server + database server. They worked in some industry configuration based on soft starters, there were about 15-20 users in the system, and the system performance was in principle suitable for everyone.
Time passed, everything worked more or less stable. But Europe imposed sanctions against Russia, as a result of which the Russians began to buy mainly domestic products, and things at this company went uphill. The number of users has grown to 50-60 people, a new branch has opened, and so has the workflow. And now the current server has ceased to cope with the sharply increased load, and 1C started, as they say, "slow down". At peak hours, documents were held for several minutes, blocking errors rained down, forms opened for a long time, well, and the whole other bunch of related services. The local system administrator dismissed all the problems, saying, "This is your 1C, you have to figure it out." We have repeatedly suggested conducting an audit of the system for performance, but this didn’t reach the audit itself.
Well, I sat down and wrote a rather voluminous letter stating that it is necessary to separate the roles of the terminal server and the application server from the DBMS (which, in principle, we have already said repeatedly before). I wrote about DFSS on terminal servers, about Shared Memory, threw links to authoritative sources and even offered some hardware options. This letter reached the powers that be in the company, went back to the IT department with the “Execute” resolutions, and the ice generally broke.
After some time, the admin sends me the IP address of the new server and login credentials. He says that MS SQL and the components of the 1C server are deployed there, and you need to transfer the databases, but so far only to the DBMS server, as there were some problems with the 1C keys.
Indeed, all the services came in, the server is not very powerful, okay, I think it's better than nothing. I’ll transfer the databases so far to at least somehow relieve the current server. In the negotiated time, he performed all the transfers, but the situation did not change - all the same performance problems. Strange, of course, well, let's register the bases in the 1C cluster, we'll see.
It takes several days, the keys have not been transferred. I’m interested in what the problem is, everything seems to be simple there - I removed it from one server, stuck it in another, installed the driver and it’s ready. The admin responds, says something about port forwarding, a virtual server, and more.
Hmm ... a virtual server? It seems that there has never been any virtualization and they didn’t exist ... I recall a rather well-known problem with the inability to forward the 1C server key to the virtual machine on Hyper-V in Windows Server 2008. And here some suspicions begin to build up ... I
open the server manager - Roles - a new one role is Hyper-V. I go to the Hyper-V manager, I see one virtual machine, I’m connecting ... And really ... Our new database server ...
So what? The instructions of the authorities and my recommendations are fulfilled, the roles are separated. The task can be closed.
After some time, the crisis now happened, the new branch had to be closed, the load decreased, the system performance became more or less tolerable.
Well, of course, they could not forward the server key to the virtual machine. As a result, everything was as it was and left: the terminal server + 1C cluster on the physical machine, the database server in the same place in the virtual one.
And okay, would it be some sort of Sharashkin office. So no. A well-known company, the products of which you probably know and have seen in the corresponding departments of all kinds of Ribbons and Auchanov.
Hard Drive Vacation Schedule
One large holding company with ambitious plans
First of all, it is necessary to deploy combat and test bases. The developer received the data for the connection, logs on to the server, sees the installed MS SQL, server 1C, sees 2 logical drives: a 250 gigabyte C drive and a 1 terabyte D drive. Well, “C” is a system, “D” for data, the developer logically decides and deploys all the databases there. I even set up maintenance plans, including backups, just in case (although we are not responsible for this). True backups took shape here on the "D". In the future, it was planned to reconfigure already on some separate network resource.
The project was launched, consultants conducted training on how to work in the new system, leftovers were transferred, some small point improvements were made and users began to work already in the new information base.
Everything went well until one morning on Monday it was discovered that the database drive had disappeared. There is simply no “D” on the server and that’s it.
Further investigation revealed this: in fact, this “server” was the working computer of the local system administrator. True, the server OS was still on it. A personal USB disk of this admin was inserted into the server. And the administrator went on vacation taking with him his own screw, with the aim of pumping films on it on the road.
Thank God, he did not manage to delete the database files and managed to restore the working database.
It is noteworthy that everyone, in general, was satisfied with the performance of the system located on the USB-drive. Nobody complained about any unsatisfactory work of 1C. It was already later that the megaproject began at the holding to transfer all information databases to a single centralized platform with super-servers, storage for a million + rubles, sophisticated hypervisors and unbearable 1C brakes in all branches.
But this is a completely different story ...