JPH0546502A - Device controller containing path block-up function per port - Google Patents

Abstract

(57) [Summary] [Purpose] The purpose of this device controller is to prevent the failure of one host from affecting other hosts. [Configuration] An error between a channel adapter 4 or 5 having a plurality of ports for establishing separate paths between channels of different hosts 1 and 2, and an error between the host 1 and the channel adapter 4 of the established path In a device controller provided with an error register 7 for holding data, it is recognized based on the content of the error register 7 whether a generated error is a higher-order error or a lower-order error. Host # 2 for port # 1 is closed by closing port # 0 only and disconnecting the established path.
Access is allowed, and when there is a lower level error, the channel adapter 4 is made inoperable and a function is provided to disconnect the paths extended to all ports # 0 and # 1.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device controller for controlling a DASD or the like of a peripheral storage device of a computer system, and more particularly to a channel having a plurality of ports for establishing separate paths with channels of different hosts. The present invention relates to a device controller including an adapter and an error storage unit that holds error data between a host of an established path and a channel adapter.

Generally, as a computer system grows in size, one device controller is shared by a plurality of hosts, and a channel adapter in the device controller is provided with a plurality of ports. The ports are associated with the hosts. A path is provided to multiplex the systems, and at this time, it is required to prevent a failure of one system from affecting another system, and the present invention responds to such a request. It is for.

[0003]

2. Description of the Related Art A general multiplexed device controller system will be described with reference to FIG. In the figure, 41 to 42 are hosts, 43 is a device controller,
44 to 45 are devices (DASD), 46 are channels of host 41 (# 0 to # 3), 47 are channels of host 42 (# 0 to ##).
3), 48 to 51 are channel adapters, 52 to 55 are device adapters, and 56 is a common bus. It should be noted that the actual numbers of hosts and devices are arbitrary according to the scale of the system.

Here, each channel adapter 48-51 is # 0.
2 ports # 1 and # 1, each port of each channel adapter has a path extending from one channel of each of the hosts 41 and 42, and has a symmetrical path arrangement as a whole. Each of the adapters 52 and 54 has a path with the device 44, and each of the device adapters 53 and 55 has a path with the device 45.

As described above, the device controller 43 and the hosts 41 to 42 are multiplexed and the device controller 43 and the devices 44 to 45 are multiplexed. For example, when the host 41 accesses the device 44. Can be performed via any one of the channel access 48 to 51 and one of the device adapters 52 to 55. Even if some paths are blocked, access is performed by using other paths. The path is arranged so that you can do it.

On the other hand, when the device controller detects an error between the operating host and the channel adapter,
An error bit is added to the status to report to the host, and the host that receives this counts the number of errors, and when the number exceeds a certain number within a certain period of time, each of its own Device controller 43 with all channel adapters 48-51 spanning channels and paths
Processing such as disconnecting from.

[0007]

As described above, in the conventional device controller, when an error occurs due to a failure with the host 41, for example, all the channel adapters 48 to 48 are instructed by the host 41. 51 is inactive, so when this failure occurs on the host 41 side, that is, the other host 42 and the channel adapter 48-51
Even when there is no fault between the host and the device, it is impossible to access the device controller from the host 42 via the path, and the system on the host side where no fault has occurred is stopped. there were.

Therefore, in the present invention, based on the contents of the error register provided in the device controller, the generated error is based on the fault of the host device (host) or the fault of the channel adapter. If the result is the former, it closes only the port connected to the higher-level device and secures the accessible state for other ports, and the failure of one host does not affect other host systems. The purpose is to do so.

[0009]

FIG. 1 illustrates the principle of the present invention. In the figure, 1 and 2 are hosts,
Each has a plurality of channels, and data transfer with the device controller is executed through the channels. A device controller 3 includes a plurality of channel adapters and is connected with a device such as DASD as a lower device. Channel adapters 4 and 5 constitute an interface with a channel of a host, each of which is provided with a plurality of ports having a path with the channel of each host, and controls the port. Reference numeral 6 denotes an MPU, which detects a data transfer from each host and operates one of the ports # 0 and # 1 to establish a path with the host which starts the data transfer, and an error content. The path is separated according to. 7 is
This is an error register (error storage unit), and sets error data indicating the error occurrence state between the host and the channel adapter via the established path, that is, the path that is actually transferring data. Has been done. 8 and 9
Is a device, and DASD or the like is used as an external storage device.

The procedure for disconnecting the path in the device controller when an error occurs is as follows. That is, the transfer data set in the register of the channel adapter.
The host determines which host the transfer is from, establishes a path with that host, and proceeds to the next step.
Here, it is assumed that a path is established between the channel # 0 of the host 1 and the port # 0 of the channel adapter 4. The contents of the error register 7 are checked and the process proceeds to the next step. Determine if there are any errors and click YES.
In case of, proceed to the next step, and in case of “NO”, return to step. Whether the error is a higher-level error generated on the host 1 side from port # 0 of the channel adapter 4 or
It is determined whether the error is a low-order error that has occurred inside the channel adapter 4 other than 0, and if it is the "former", proceed to the next step, and if it is the "latter", proceed to the step. A process for closing only the port # 0 of the channel adapter 4 is performed to disconnect the path established with the channel # 0 of the host 1. At this time, since the port # 1 of the channel adapter 4 is not closed, access from the channel # 0 of the host 2 to the port # 1 is possible. The channel adapter 4 itself is deactivated and disconnected. At this time, the port # 1 of the channel adapter 4 is also closed. The disconnection processing is performed according to the content of the error by such a procedure.

[0011]

According to the present invention, the contents of the error register in which the error occurs in the exchange of data in the system from the host to the channel adapter are checked in this way, and When the place of occurrence is on the host side (including the port of the established path), the process of closing only that port is performed and other ports are accessible.

Further, unlike a temporary failure that can be recovered immediately, a failure of disconnecting a currently used path often occurs repeatedly in a relatively short time. It is general to make a determination as to whether or not an error has occurred a certain number of times or more in a certain time.

[0013]

Embodiments of the present invention will be described with reference to FIGS. 2 is an explanatory diagram showing the inside of the device controller, and FIG. 3 is an explanatory diagram showing the procedure from the establishment of the path to the disconnection by the application program.

In FIG. 2, 11 is a channel adapter and 12
Is a common bus, 13 is a device adapter, 14 is a service adapter that has a timer inside to monitor the device controller, and 15 is an RM (Resou) that controls each functional module (adapter) in the device controller to perform overall management.
rce Manager), 16 to 17 are port of channel adapter, 18
Is an application program, 19 is a register group, 20 ~
Reference numeral 21 is an access recognition register, 22 is an error register indicating an error occurrence state between the host and the channel adapter, and 23 is a micro disk.

The procedure from the path establishment to the disconnection by the application program is as follows. That is, (1) For example, when a signal from the host is sent to port 16, the access recognition register corresponding to port 16
Set a bit "1" as shown in 20 and proceed to the next step. (2) By detecting the bit "1" of the register 20 by periodically checking the access recognition registers 20 and 21, it is recognized that there is an access request from the host to port 16, and the next step Proceed to. (3) Determine whether the content of the data sent from the host is normal, and if “YES”, proceed to the next step and click “N
In the case of "O", the processing ends. (4) The RM 15 is inquired as to whether or not data transfer processing with the host may be performed via the port 16, and the process proceeds to the next step. (5) Judge whether the reply from RM15 is "OK". If "YES", proceed to the next step, and if "NO", cancel the process. (6) Host via port 16 Path to the channel adapter 11 and proceed to the next step. (7) When an error occurs more than a certain number of times within a certain time by checking the error register 22 regularly, whether the error is a higher level error that occurred on the host side including port 16 or not It is confirmed whether the error is a lower level error that has occurred in the channel adapter 11, and the process proceeds to the next step. (8) When it is a higher error, only port 16 is closed and the path connected to that port is disconnected, and when it is a lower error, the entire channel adapter 11 is disabled and ports 16, 17 Disconnect both of the paths that are connected to each. The path is separated according to the contents of the error by the following procedure.

Here, the service adapter 14 can also execute error detection and path disconnection processing according to the error content. That is, the service adapter 14 identifies the time and place of occurrence of an error that has occurred in the device controller, and the micro disk 23
When an error occurs more than a certain number of times within a certain period of time, if the error that is the cause of the error is on the upper device (used port and host) side, It is determined whether it is on the device side (device controller excluding the ports used).

Next, when the result of the determination is the former, the service adapter 14 instructs the channel adapter 11 to close only the port in which many errors occur, and the channel adapter 11 that has received this instruction It closes the port artificially by reporting busy to the specified port, and does not respond to subsequent access from the host, and in the latter case, the module that has failed, for example, By degrading the entire channel adapter 11, it becomes inoperable and prompts the host side of the host device to close the path.

[0018]

As described above, according to the present invention, in a system in which a plurality of hosts have paths to the same channel adapter in a device controller, when an error occurs frequently between a host and a port of an established path, Check whether the occurrence location is on the host (higher-level device) side that includes the relevant port, or in the channel adapter excluding the relevant port, and if it is determined that the fault is due to the higher-level device, then only the operating port. Even if the corresponding path is disconnected due to the failure of a higher-level device, the path between the other high-level device and the channel adapter remains accessible, so that However, it does not affect other host system using the same channel adapter.

[Brief description of drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is an explanatory diagram showing the inside of a device controller according to the present invention.

FIG. 3 is an explanatory diagram showing a procedure from establishment of a path to disconnection by an application program according to the present invention.

FIG. 4 is an explanatory diagram showing a general multiplexed device controller system.

[Explanation of symbols]

 In FIG. 1, 1, 2 ... host 3 ... device controller 4,5 ... channel adapter 6 ... MPU 7 ... error register (error storage unit) 8, 9 ... device (DASD )

Claims (2)

[Claims] 1. A channel adapter having a plurality of ports for establishing separate paths with channels of different hosts, and an error storage for holding error data between the host and the channel adapter of the established path. In the device controller including a section, based on the contents of the error storage section,
Recognize whether it is a higher-level error due to a failure on the host side that includes the port of the established path, or a lower-level error due to a failure in the channel adapter other than the relevant port, and established when it is a higher-level error It has a function to close only the port of the path and disconnect the path, and when there is a lower error, make the channel adapter inoperable and disconnect the path set to all ports of the relevant channel adapter. At this time, a device controller having a path blocking function for each port, which allows a host not having a path established to remain accessible to a host host. 2. The device controller having a path blocking function for each port according to claim 1, wherein an error register provided in the channel adapter is used as the error storage unit.