Skip to contentSkip to conversation

sg3_utils: udev rules: restrict use of ambiguous device IDs

RFC PATCH In review [email protected] 2 participants 6 messages Mar 27, 2023 – Mar 28, 2023

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover sg3_utils: udev rules: restrict use of ambiguous device IDs

From: Martin Wilck <[email protected]>

Most modern SCSI devices provide VPD page 83 with at least one highly reliable device identifier, like NAA Registered Extended or EUI-64, or the ata-id identifier. Other device identifier types have shown to be less reliable and possibly ambiguous. Ambiguity in particular is a problem with multipath-tools, which may group unrelated devices together in a multipath map, causing possible data corruption.

The device identifiers are used in two independent ways by the udev rules:
a) to set ID_SERIAL for subsystems like multipath, and b) to create /dev/disk/by-id/scsi-... symlinks. Our udev rules have traditionally created symlinks for every device identifier obtained from either VPD 83 or 80. This may cause issues, especially on large installments with storage devices that exhibit the same identifier for many logical units. At the same time, these symlinks are rarely used.

Avoid using unreliable identifiers for setting ID_SERIAL, and don't create symlinks for these identifiers. Add a configuration method that allows users to easily re-enable these methods and symlinks if they need to (this might be the case on systems with legacy devices that are referenced in /etc/crypttab, lvm.conf, or the like). This is done by introducing environment variables .SCSI_ID_SERIAL_SRC and .SCSI_ID_SYMLINK_SRC, to control use of device identifiers for determining ID_SERIAL and for creating symlinks, respectively. Both variables can contain the letters "T", "L", "V", and "S" to enable T10-vendor ID, NAA local ID, vendor-specific ID, and VPD 80 based ID, respectively.

Distributions can change the defaults for these environment variables to provide backward compatibility for their users, while offering users an easy way to change the settings.

I'm sending this as RFC, because I expect that not everyone will agree which identifiers should be enabled by default.

Martin Wilck (3):
55-scsi-sg3_id.rules: don't set unreliable device ID by default 58-scsi-sg3_symlink.rules: don't create extra by-id symlinks by
default
udev: add 00-scsi-sg3_config.rules for user configuration

 Makefile.am                       |  1 +
 scripts/00-scsi-sg3_config.rules  | 23 ++++++++++++++
 scripts/55-scsi-sg3_id.rules      | 53 ++++++++++++++++++++++++++++---
 scripts/58-scsi-sg3_symlink.rules | 46 +++++++++++++++++++++------
 4 files changed, 109 insertions(+), 14 deletions(-)
 create mode 100644 scripts/00-scsi-sg3_config.rules
Signature · 6 lines
2.39.2

--
dm-devel mailing list
[email protected]
https://listman.redhat.com/mailman/listinfo/dm-devel
#1 lore raw

PATCH 1/3 55-scsi-sg3_id.rules: don't set unreliable device ID by default

From: Martin Wilck <[email protected]>

Some of the methods to determine the SCSI device ID are known to be unreliable and, and possibly cause dm-multipath to falsely detect a multipath setup and cause data corruption. These methods will only be used if the known reliable device IDs (like NAA registered) are unavailable.

Don't apply these methods by default. Make the methods to use configurable instead. Configuration is done by setting the udev property ".SCSI_ID_SERIAL_SRC" to any combination of the letters T, L, V, S:

T: T10 vendor ID ("1...")
L: NAA local ("33...")
V: vendor-specific ("0...")
S: vendor/model/serial number ("S...")

The ordering of the letters in .SCSI_ID_SERIAL_SRC doesn't matter, the precedence is always T, L, V, S.

Regardless of using these properties for ID_SERIAL, we can infer that if any of these properties is set, it's correct to set ID_BUS="scsi".

---
 scripts/55-scsi-sg3_id.rules | 53 +++++++++++++++++++++++++++++++++---
 1 file changed, 49 insertions(+), 4 deletions(-)
scripts/55-scsi-sg3_id.rules +49 -4
index 453210b..33b2ad3 100644
@@ -99,11 +99,56 @@ ENV{ID_SERIAL}!="?*", ENV{SCSI_IDENT_LUN_NAA_REG}=="?*", ENV{ID_BUS}="scsi", ENV
 ENV{ID_SERIAL}!="?*", ENV{SCSI_IDENT_LUN_NAA_EXT}=="?*", ENV{ID_BUS}="scsi", ENV{ID_SERIAL}="3$env{SCSI_IDENT_LUN_NAA_EXT}", ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_NAA_EXT}"
 ENV{ID_SERIAL}!="?*", ENV{SCSI_IDENT_LUN_EUI64}=="?*", ENV{ID_BUS}="scsi", ENV{ID_SERIAL}="2$env{SCSI_IDENT_LUN_EUI64}", ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_EUI64}"
 ENV{ID_SERIAL}!="?*", ENV{SCSI_IDENT_LUN_NAME}=="?*", ENV{ID_BUS}="scsi", ENV{ID_SERIAL}="8$env{SCSI_IDENT_LUN_NAME}", ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_NAME}"
-ENV{ID_SERIAL}!="?*", ENV{SCSI_IDENT_LUN_T10}=="?*", ENV{ID_BUS}="scsi", ENV{ID_SERIAL}="1$env{SCSI_IDENT_LUN_T10}", ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_T10}"
-ENV{ID_SERIAL}!="?*", ENV{SCSI_IDENT_LUN_NAA_LOCAL}=="?*", ENV{ID_BUS}="scsi", ENV{ID_SERIAL}="3$env{SCSI_IDENT_LUN_NAA_LOCAL}", ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_NAA_LOCAL}"
-ENV{ID_SERIAL}!="?*", ENV{SCSI_IDENT_LUN_VENDOR}=="?*", ENV{ID_BUS}="scsi", ENV{ID_SERIAL}="0$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_LUN_VENDOR}", ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_VENDOR}"
-ENV{ID_SERIAL}!="?*", ENV{SCSI_IDENT_SERIAL}=="?*", ENV{ID_BUS}="scsi", ENV{ID_SERIAL}="S$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_SERIAL}", ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_SERIAL}"
 
 # Compat ID_SCSI_SERIAL setting
 ENV{ID_SCSI_SERIAL}!="?*", ENV{SCSI_IDENT_SERIAL}=="?*", ENV{ID_SCSI_SERIAL}="$env{SCSI_IDENT_SERIAL}"
+
+# Enable or disable possibly ambiguous SCSI device ID sources for setting ID_SERIAL
+#
+# .SCSI_ID_SERIAL_SRC can be any combination of the characters "TLVS":
+#   T: T10 vendor ID ("1...") from VPD 0x83
+#   L: NAA local ("33...") from VPD 0x83
+#   V: vendor-specific ("0...") from VPD 0x83
+#   S: vendor/model/serial number ("S...") from VPD 0x80
+# The ordering of the letters doesn't matter, the precedence is always T, L, V, S
+# NAA (except "local") and EUI-64 IDs (see below), as well as ATA IDs, always take precedence over
+# the sources configured here.
+#
+# This only needs to be changed if there are legacy SCSI devices that don't provide any reliable
+# device identifiers, and some subsystem like multipath requires that ID_SERIAL is set.
+# Be aware that multipath actually needs unique identifiers, though.
+# Using ambiguous identifiers for ID_SERIAL can cause data corruption with multipath.
+#
+# To configure this, add an early rule (e.g. /etc/udev/rules.d/00-scsi-serial.rules) e.g. like this:
+# ACTION!="remove", KERNEL=="sd*|sr*|st*|nst*|cciss*", ENV{.SCSI_ID_SERIAL_SRC}="TLVS"
+#
+# By default, only T10 vendor ID is allowed.
+ENV{.SCSI_ID_SERIAL_SRC}!="?*", ENV{.SCSI_ID_SERIAL_SRC}="T"
+
+ENV{ID_SERIAL}=="?*", GOTO="sg3_utils_id_end"
+ENV{SCSI_IDENT_LUN_T10}=="?*", ENV{ID_BUS}="scsi"
+ENV{SCSI_IDENT_LUN_T10}=="?*", ENV{.SCSI_ID_SERIAL_SRC}=="*T*", \
+    ENV{ID_SERIAL}="1$env{SCSI_IDENT_LUN_T10}", \
+    ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_T10}"
+
+ENV{ID_SERIAL}=="?*", GOTO="sg3_utils_id_end"
+ENV{SCSI_IDENT_LUN_NAA_LOCAL}=="?*", ENV{ID_BUS}="scsi"
+ENV{SCSI_IDENT_LUN_NAA_LOCAL}=="?*", ENV{.SCSI_ID_SERIAL_SRC}=="*L*", \
+    ENV{ID_SERIAL}="3$env{SCSI_IDENT_LUN_NAA_LOCAL}", \
+    ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_NAA_LOCAL}"
+
+ENV{ID_SERIAL}=="?*", GOTO="sg3_utils_id_end"
+ENV{SCSI_IDENT_LUN_VENDOR}=="?*", ENV{ID_BUS}="scsi"
+ENV{SCSI_IDENT_LUN_VENDOR}=="?*", ENV{.SCSI_ID_SERIAL_SRC}=="*V*", \
+    ENV{ID_SERIAL}="0$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_LUN_VENDOR}", \
+    ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_LUN_VENDOR}"
+
+ENV{ID_SERIAL}=="?*", GOTO="sg3_utils_id_end"
+ENV{SCSI_IDENT_SERIAL}=="?*", ENV{ID_BUS}="scsi"
+ENV{SCSI_IDENT_SERIAL}=="?*", ENV{.SCSI_ID_SERIAL_SRC}=="*S*", \
+    ENV{ID_SERIAL}="S$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_SERIAL}", \
+    ENV{ID_SERIAL_SHORT}="$env{SCSI_IDENT_SERIAL}"
+
 LABEL="sg3_utils_id_end"
+ENV{ID_SERIAL}!="?*", ENV{DEVTYPE}=="disk", \
+    PROGRAM="/bin/logger -t 55-scsi-sg3_id.rules -p daemon.warning \"WARNING: SCSI device %k has no device ID, consider changing .SCSI_ID_SERIAL_SRC in 00-scsi-sg3_config.rules\""
Signature · 6 lines
2.39.2

--
dm-devel mailing list
[email protected]
https://listman.redhat.com/mailman/listinfo/dm-devel
#2 lore raw

PATCH 2/3 58-scsi-sg3_symlink.rules: don't create extra by-id symlinks by default

In reply to [email protected]

From: Martin Wilck <[email protected]>

The current code will set device symlinks for every identifier obtained from VPD pages 0x83 and 0x80. This is seldom useful. Device IDs shouldn't be used for identifying file systems or high-level objects such as LVs; it is much more useful to identify these by their respective UUIDs. Those subsystems that need device IDs, such as LVM, multipath, dm-crypt etc, just need one identifier. For these use cases, a single symlink is sufficient, and it should be the most reliable one as selected by the ID_SERIAL logic in 55-scsi-sg3_id.rules.

On the other hand, especially on large configurations, unreliable and ambiguous device identifiers can cause trouble when many devices claim the same symlink. udev's attempts to determine the highest-priority contender for a given symlink may be very resource-intensive and slow, especially during boot, when lots of uevents for similar devices have to be processed in parallel. This can cause udev workers to be killed, and in the worst case, boot failure.

Avoid these issues by not creating possibly ambiguous /dev/disk/by-id symlinks any more by default. Users can modify the configuration by setting the types of symlinks to create in the environment variable .SCSI_SYMLINK_SRC, which can be a combination of the letters T, L, V, S:

T: T10 vendor ID ("1...") from VPD 0x83
L: NAA local ("33...") from VPD 0x83
V: vendor-specific ("0...") from VPD 0x83
S: vendor/model/serial number ("S...") from VPD 0x80

In practice, modifying this should only be necessary if legacy devices that don't provide any reliable identifiers are used as targets for dm-crypt or LVM.

---
 scripts/58-scsi-sg3_symlink.rules | 50 +++++++++++++++++++++++++------
 1 file changed, 41 insertions(+), 9 deletions(-)
scripts/58-scsi-sg3_symlink.rules +41 -9
index fe6b000..99fdc23 100644
@@ -5,10 +5,36 @@ ACTION=="remove", GOTO="sg3_utils_symlink_end"
 SUBSYSTEM!="block", GOTO="sg3_utils_symlink_end"
 ENV{UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG}=="1", GOTO="sg3_utils_symlink_end"
 
-# Select which identifier to use per default
+# Enable or disable possibly ambiguous SCSI device symlinks under /dev/disk/by-id
+#
+# .SCSI_SYMLINK_SRC can be any combination of the letter "TLVS":
+#   T: T10 vendor ID ("1...") from VPD 0x83
+#   L: NAA local ("33...") from VPD 0x83
+#   V: vendor-specific ("0...") from VPD 0x83
+#   S: vendor/model/serial number ("S...") from VPD 0x80
+# Symlinks will be created for every letter included in .SCSI_SYMLINK_SRC.
+# Symlinks for NAA (except "local") and EUI-64 IDs (see below) are always created.
+#
+# NOTE: The default rules in 60-persistent-storage.rules create a symlink
+# "ENV{ID_BUS}-ENV{ID_SERIAL}" symlink anyway, where ID_BUS is "scsi", "ata", "usb", or "cciss".
+# ID_SERIAL is set in 55-scsi-sg3_id.rules from the least ambiguous device identifier.
+# The symlinks created by this file are created *in addition* to the default symlink.
+#
+# This only needs to be changed if some subsystem, like dm-crypt or LVM, depends on the
+# additional symlinks being present for device identification.
+#
+# To configure the behavior, add an early rule (e.g. /etc/udev/rules.d/00-scsi-serial.rules)
+# like this:
+# ACTION!="remove", KERNEL=="sd*|sr*|st*|nst*|cciss*", ENV{.SCSI_SYMLINK_SRC}="TS"
+#
+# By default, no possibly ambiguous additional symlinks will be created.
+ENV{.SCSI_SYMLINK_SRC}!="?*", ENV{.SCSI_SYMLINK_SRC}=""
+
 # 0: vpd page 0x80 identifier
-ENV{SCSI_IDENT_SERIAL}=="?*", ENV{DEVTYPE}=="disk", SYMLINK+="disk/by-id/scsi-S$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_SERIAL}"
-ENV{SCSI_IDENT_SERIAL}=="?*", ENV{DEVTYPE}=="partition", SYMLINK+="disk/by-id/scsi-S$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_SERIAL}-part%n"
+ENV{.SCSI_SYMLINK_SRC}=="*S*", ENV{SCSI_IDENT_SERIAL}=="?*", ENV{DEVTYPE}=="disk", \
+    SYMLINK+="disk/by-id/scsi-S$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_SERIAL}"
+ENV{.SCSI_SYMLINK_SRC}=="*S*", ENV{SCSI_IDENT_SERIAL}=="?*", ENV{DEVTYPE}=="partition", \
+    SYMLINK+="disk/by-id/scsi-S$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_SERIAL}-part%n"
 # NAA identifier (prefix 3)
 # 1: IEEE Registered Extended first
 ENV{SCSI_IDENT_LUN_NAA_REGEXT}=="?*", ENV{DEVTYPE}=="disk", SYMLINK+="disk/by-id/scsi-3$env{SCSI_IDENT_LUN_NAA_REGEXT}"
@@ -26,13 +52,19 @@ ENV{SCSI_IDENT_LUN_EUI64}=="?*", ENV{DEVTYPE}=="partition", SYMLINK+="disk/by-id
 ENV{SCSI_IDENT_LUN_NAME}=="?*", ENV{DEVTYPE}=="disk", SYMLINK+="disk/by-id/scsi-8$env{SCSI_IDENT_LUN_NAME}"
 ENV{SCSI_IDENT_LUN_NAME}=="?*", ENV{DEVTYPE}=="partition", SYMLINK+="disk/by-id/scsi-8$env{SCSI_IDENT_LUN_NAME}-part%n"
 # 6: T10 Vendor identifier (prefix 1)
-ENV{SCSI_IDENT_LUN_T10}=="?*", ENV{DEVTYPE}=="disk", SYMLINK+="disk/by-id/scsi-1$env{SCSI_IDENT_LUN_T10}"
-ENV{SCSI_IDENT_LUN_T10}=="?*", ENV{DEVTYPE}=="partition", SYMLINK+="disk/by-id/scsi-1$env{SCSI_IDENT_LUN_T10}-part%n"
+ENV{.SCSI_SYMLINK_SRC}=="*T*", ENV{SCSI_IDENT_LUN_T10}=="?*", ENV{DEVTYPE}=="disk", \
+    SYMLINK+="disk/by-id/scsi-1$env{SCSI_IDENT_LUN_T10}"
+ENV{.SCSI_SYMLINK_SRC}=="*T*", ENV{SCSI_IDENT_LUN_T10}=="?*", ENV{DEVTYPE}=="partition", \
+    SYMLINK+="disk/by-id/scsi-1$env{SCSI_IDENT_LUN_T10}-part%n"
 # 7: IEEE Locally assigned
-ENV{SCSI_IDENT_LUN_NAA_LOCAL}=="?*", ENV{DEVTYPE}=="disk", SYMLINK+="disk/by-id/scsi-3$env{SCSI_IDENT_LUN_NAA_LOCAL}"
-ENV{SCSI_IDENT_LUN_NAA_LOCAL}=="?*", ENV{DEVTYPE}=="partition", SYMLINK+="disk/by-id/scsi-3$env{SCSI_IDENT_LUN_NAA_LOCAL}-part%n"
+ENV{.SCSI_SYMLINK_SRC}=="*L*", ENV{SCSI_IDENT_LUN_NAA_LOCAL}=="?*", ENV{DEVTYPE}=="disk", \
+    SYMLINK+="disk/by-id/scsi-3$env{SCSI_IDENT_LUN_NAA_LOCAL}"
+ENV{.SCSI_SYMLINK_SRC}=="*L*", ENV{SCSI_IDENT_LUN_NAA_LOCAL}=="?*", ENV{DEVTYPE}=="partition", \
+    SYMLINK+="disk/by-id/scsi-3$env{SCSI_IDENT_LUN_NAA_LOCAL}-part%n"
 # 8: Vendor-specific identifier (prefix 0)
-ENV{SCSI_IDENT_LUN_VENDOR}=="?*", ENV{DEVTYPE}=="disk", SYMLINK+="disk/by-id/scsi-0$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_LUN_VENDOR}"
-ENV{SCSI_IDENT_LUN_VENDOR}=="?*", ENV{DEVTYPE}=="partition", SYMLINK+="disk/by-id/scsi-0$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_LUN_VENDOR}-part%n"
+ENV{.SCSI_SYMLINK_SRC}=="*V*", ENV{SCSI_IDENT_LUN_VENDOR}=="?*", ENV{DEVTYPE}=="disk", \
+    SYMLINK+="disk/by-id/scsi-0$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_LUN_VENDOR}"
+ENV{.SCSI_SYMLINK_SRC}=="*V*", ENV{SCSI_IDENT_LUN_VENDOR}=="?*", ENV{DEVTYPE}=="partition", \
+    SYMLINK+="disk/by-id/scsi-0$env{SCSI_VENDOR}_$env{SCSI_MODEL}_$env{SCSI_IDENT_LUN_VENDOR}-part%n"
 
 LABEL="sg3_utils_symlink_end"
Signature · 6 lines
2.39.2

--
dm-devel mailing list
[email protected]
https://listman.redhat.com/mailman/listinfo/dm-devel
#3 lore raw

PATCH 3/3 udev: add 00-scsi-sg3_config.rules for user configuration

In reply to [email protected]

From: Martin Wilck <[email protected]>

Add a dedicated early rules file to simplify configuring the behavior of the SCSI udev rules shipped with sg3_utils.

---
 Makefile.am                       |  1 +
 scripts/00-scsi-sg3_config.rules  | 23 +++++++++++++++++++++++
 scripts/55-scsi-sg3_id.rules      |  6 ++----
 scripts/58-scsi-sg3_symlink.rules |  6 +-----
 4 files changed, 27 insertions(+), 9 deletions(-)
 create mode 100644 scripts/00-scsi-sg3_config.rules
Makefile.am +1 -0
index 240acbe..ce7ee71 100644
@@ -123,6 +123,7 @@ EXTRA_DIST += \
 	inhex/z_act_query.hex
 
 EXTRA_DIST += \
+	scripts/00-scsi-sg3_config.rules \
 	scripts/40-usb-blacklist.rules \
 	scripts/54-before-scsi-sg3_id.rules \
 	scripts/55-scsi-sg3_id.rules \
scripts/00-scsi-sg3_config.rules new file +23 -0
new file mode 100644
index 0000000..43d6f16
@@ -0,0 +1,23 @@
+# Configuration for SCSI device identification
+
+# To apply changes, copy this file to /etc/udev/rules.d and edit to suit your needs.
+# DO NOT EDIT THIS FILE IN PLACE!
+
+ACTION!="add|change", GOTO="scsi_identify_end"
+SUBSYSTEMS=="scsi", GOTO="scsi_identify"
+GOTO="scsi_identify_end"
+LABEL="scsi_identify"
+
+# Set ID_SCSI_INQUIRY to 0 to force running "sg_inq" for obtaining device IDs
+# from SCSI VPDs, rather than looking them up in sysfs (not recommended).
+ENV{ID_SCSI_INQUIRY}=""
+
+# Set enabled unreliable sources for setting the ID_SERIAL property.
+# See 55-scsi-sg3_id.rules for detailed documentation.
+ENV{.SCSI_ID_SERIAL_SRC}="T"
+
+# Set enabled unreliable sources for creating additional /dev/disk/by-id/scsi* symlinks.
+# See 58-scsi-sg3_symlink.rules for detailed documentation.
+ENV{.SCSI_SYMLINK_SRC}=""
+
+LABEL="scsi_identify_end"
scripts/55-scsi-sg3_id.rules +2 -4
index 33b2ad3..5e9732d 100644
@@ -36,7 +36,7 @@ KERNEL!="sd*[!0-9]|sr*", GOTO="sg3_utils_id_end"
 # have scanned for VPD pages, so if the vpd page attribute is not
 # present it is not supported (or deemed unsafe to access).
 # Hence we can skip the call to sg_inq and avoid I/O altogether.
-# Set 'ID_SCSI_INQUIRY=0' in an earlier udev rule if the kernel
+# Set ENV{ID_SCSI_INQUIRY}="0" in 00-scsi-sg3_config.rules if the kernel
 # fails to scan VPD pages correctly; the rules will then fall
 # back to calling sg_vpd directly.
 LABEL="scsi_inquiry"
@@ -119,9 +119,7 @@ ENV{ID_SCSI_SERIAL}!="?*", ENV{SCSI_IDENT_SERIAL}=="?*", ENV{ID_SCSI_SERIAL}="$e
 # Be aware that multipath actually needs unique identifiers, though.
 # Using ambiguous identifiers for ID_SERIAL can cause data corruption with multipath.
 #
-# To configure this, add an early rule (e.g. /etc/udev/rules.d/00-scsi-serial.rules) e.g. like this:
-# ACTION!="remove", KERNEL=="sd*|sr*|st*|nst*|cciss*", ENV{.SCSI_ID_SERIAL_SRC}="TLVS"
-#
+# To configure the behavior, set ENV{.SCSI_ID_SERIAL_SRC} in 00-scsi-sg3_config.rules.
 # By default, only T10 vendor ID is allowed.
 ENV{.SCSI_ID_SERIAL_SRC}!="?*", ENV{.SCSI_ID_SERIAL_SRC}="T"
 
scripts/58-scsi-sg3_symlink.rules +1 -5
index 99fdc23..dfe8f77 100644
@@ -23,12 +23,8 @@ ENV{UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG}=="1", GOTO="sg3_utils_symlink_en
 # This only needs to be changed if some subsystem, like dm-crypt or LVM, depends on the
 # additional symlinks being present for device identification.
 #
-# To configure the behavior, add an early rule (e.g. /etc/udev/rules.d/00-scsi-serial.rules)
-# like this:
-# ACTION!="remove", KERNEL=="sd*|sr*|st*|nst*|cciss*", ENV{.SCSI_SYMLINK_SRC}="TS"
-#
+# To configure the behavior, set ENV{.SCSI_SYMLINK_SRC} in 00-scsi-sg3_config.rules.
 # By default, no possibly ambiguous additional symlinks will be created.
-ENV{.SCSI_SYMLINK_SRC}!="?*", ENV{.SCSI_SYMLINK_SRC}=""
 
 # 0: vpd page 0x80 identifier
 ENV{.SCSI_SYMLINK_SRC}=="*S*", ENV{SCSI_IDENT_SERIAL}=="?*", ENV{DEVTYPE}=="disk", \
Signature · 6 lines
2.39.2

--
dm-devel mailing list
[email protected]
https://listman.redhat.com/mailman/listinfo/dm-devel
#4 lore raw
In reply to [email protected]
Quoted 28 lines · On 2023-03-27 09:24, [email protected] wrote:

From: Martin Wilck <[email protected]>

Most modern SCSI devices provide VPD page 83 with at least one highly reliable device identifier, like NAA Registered Extended or EUI-64, or the ata-id identifier. Other device identifier types have shown to be less reliable and possibly ambiguous. Ambiguity in particular is a problem with multipath-tools, which may group unrelated devices together in a multipath map, causing possible data corruption.

The device identifiers are used in two independent ways by the udev rules:
a) to set ID_SERIAL for subsystems like multipath, and b) to create /dev/disk/by-id/scsi-... symlinks. Our udev rules have traditionally created symlinks for every device identifier obtained from either VPD 83 or 80. This may cause issues, especially on large installments with storage devices that exhibit the same identifier for many logical units. At the same time, these symlinks are rarely used.

Avoid using unreliable identifiers for setting ID_SERIAL, and don't create symlinks for these identifiers. Add a configuration method that allows users to easily re-enable these methods and symlinks if they need to (this might be the case on systems with legacy devices that are referenced in /etc/crypttab, lvm.conf, or the like). This is done by introducing environment variables .SCSI_ID_SERIAL_SRC and .SCSI_ID_SYMLINK_SRC, to control use of device identifiers for determining ID_SERIAL and for creating symlinks, respectively. Both variables can contain the letters "T", "L", "V", and "S" to enable T10-vendor ID, NAA local ID, vendor-specific ID, and VPD 80 based ID, respectively.

Quoted 6 lines

Distributions can change the defaults for these environment variables to provide backward compatibility for their users, while offering users an easy way to change the settings.

I'm sending this as RFC, because I expect that not everyone will agree which identifiers should be enabled by default.

Lets see if anything happens. Applied as sg3_utils revision 1019 and pushed to https://github.com/doug-gilbert/sg3_utils .

Didn't see any effect on an Ubuntu 22.10 when sg3_utils deb package built and installed. No sign of 00-scsi-sg3_config.rules being placed anywhere by Ubuntu. Does Suse install those rules?

Doug Gilbert

Quoted 13 lines

Martin Wilck (3):
55-scsi-sg3_id.rules: don't set unreliable device ID by default 58-scsi-sg3_symlink.rules: don't create extra by-id symlinks by
default
udev: add 00-scsi-sg3_config.rules for user configuration

  Makefile.am                       |  1 +
  scripts/00-scsi-sg3_config.rules  | 23 ++++++++++++++
  scripts/55-scsi-sg3_id.rules      | 53 ++++++++++++++++++++++++++++---
  scripts/58-scsi-sg3_symlink.rules | 46 +++++++++++++++++++++------
  4 files changed, 109 insertions(+), 14 deletions(-)

create mode 100644 scripts/00-scsi-sg3_config.rules

#5 lore raw
Quoted 7 lines · Douglas Gilbert

Lets see if anything happens. Applied as sg3_utils revision 1019 and pushed to https://github.com/doug-gilbert/sg3_utils .

Didn't see any effect on an Ubuntu 22.10 when sg3_utils deb package built and installed. No sign of 00-scsi-sg3_config.rules being placed anywhere by Ubuntu. Does Suse install those rules?

No, not yet. That's why I sent these patches, I intend to get rid of the legacy symlinks on openSUSE with upstream's blessing.

Regards,
Martin