openldap-technical October 2017

openldap-technical@openldap.org

28 participants
75 discussions

Re: make test seg fault related to libc
by Quanah Gibson-Mount 11 Oct '17

11 Oct '17

--On Wednesday, October 11, 2017 1:02 PM -0700 Scott Classen <sclassen(a)lbl.gov> wrote: ># ulimit -a Hm, well, that doesn't appear to be it. :/ You might look at the testrun logs and see what slapd is logging when it dies as well. --Quanah -- Quanah Gibson-Mount Product Architect Symas Corporation Packaged, certified, and supported LDAP solutions powered by OpenLDAP: <http://www.symas.com>

2 1

Re: make test seg fault related to libc
by Quanah Gibson-Mount 11 Oct '17

11 Oct '17

--On Wednesday, October 11, 2017 10:44 AM -0700 Scott Classen <sclassen(a)lbl.gov> wrote: > Hello, > > I've built openldap 2.4.45 on a CentOS 7.4.1708 machine with the > following configuration: I've seen this happen when a ulimit value was unreasonably small. Check the limits on -m or -v perhaps. --Quanah -- Quanah Gibson-Mount Product Architect Symas Corporation Packaged, certified, and supported LDAP solutions powered by OpenLDAP: <http://www.symas.com>

2 1

make test seg fault related to libc
by Scott Classen 11 Oct '17

11 Oct '17

Hello, I've built openldap 2.4.45 on a CentOS 7.4.1708 machine with the following configuration: ./configure --enable-bdb=no --enable-hdb=no --enable-mdb --with-tls=openssl --enable-spasswd --enable-syslog --enable-modules --enable-cleartext --enable-overlays --enable-accesslog --enable-auditlog --with-threads --enable-shared --enable-ldap --enable-monitor --enable-deref --enable-slapd --enable-ppolicy --enable-memberof When I run make test I receive numerous segmentation faults: output from make test: > ./scripts/test028-idassert: line 252: 28923 Segmentation fault (core dumped) $SLAPD -f $CONF1 -h $URI1 -d $LVL $TIMING > $LOG1 2>&1 > >>>>> test028-idassert completed OK for mdb. > > >>>>> Starting test029-ldapglue for mdb... > running defines.sh > ### This test requires the ldap backend and glue overlay. > ### If available, and explicitly requested, it can use SASL bind; > ### note that SASL must be properly set up, and the requested > ### mechanism must be available. Define SLAPD_USE_SASL={yes|<mech>}, > ### with "yes" defaulting to DIGEST-MD5 to enable SASL authc[/authz]. > Using proxyAuthz with simple authc... > Running slapadd to build slapd database... > Starting local slapd on TCP/IP port 9011... > Starting remote slapd 1 on TCP/IP port 9012... > Starting remote slapd 2 on TCP/IP port 9013... > Using ldapsearch to check that slapd is running... > Using ldapsearch to check that slapd is running... > Using ldapsearch to check that slapd is running... > Testing ldapsearch as uid=bjorn,ou=People,dc=example,dc=com for "dc=example,dc=com"... > Filtering ldapsearch results... > Filtering original ldif used to create database... > Comparing filter output... > Testing ldapsearch as anonymous for "dc=example,dc=com"... > Filtering ldapsearch results... > Filtering original ldif used to create database... > Comparing filter output... > >>>>> Test succeeded > ./scripts/test029-ldapglue: line 222: 28992 Segmentation fault (core dumped) $SLAPD -f $CONF1 -h $URI1 -d $LVL $TIMING > $LOG1 2>&1 > ./scripts/test029-ldapglue: line 222: 28994 Segmentation fault (core dumped) $SLAPD -f $CONF2 -h $URI2 -d $LVL $TIMING > $LOG2 2>&1 > ./scripts/test029-ldapglue: line 222: 28996 Segmentation fault (core dumped) $SLAPD -f $CONF3 -h $URI3 -d $LVL $TIMING > $LOG3 2>&1 > >>>>> test029-ldapglue completed OK for mdb. In /var/log/messages: > Oct 11 09:31:42 www kernel: slapd[9039]: segfault at 1100fffff7 ip 00007febd7a2f4dc sp 00007ffe832541f8 error 4 in libc-2.17.so[7febd79af000+1b8000] > Oct 11 09:31:42 www kernel: slapd[8997]: segfault at 1100fffff7 ip 00007f48923e44dc sp 00007ffec5a22658 error 4 in libc-2.17.so[7f4892364000+1b8000] > Oct 11 09:31:42 www kernel: slapd[8975]: segfault at 1100fffff7 ip 00007f01a727d4dc sp 00007ffd9a127698 error 4 in libc-2.17.so[7f01a71fd000+1b8000] > Oct 11 09:31:42 www abrt-hook-ccpp: Process 8975 (slapd) of user 0 killed by SIGSEGV - ignoring (repeated crash) > Oct 11 09:31:42 www abrt-hook-ccpp: Process 9039 (slapd) of user 0 killed by SIGSEGV - dumping core > Oct 11 09:31:42 www abrt-hook-ccpp: Process 8997 (slapd) of user 0 killed by SIGSEGV - dumping core I don't see any obvious problems with the linked libraries: [me@here openldap-2.4.45]# ldd servers/slapd/slapd linux-vdso.so.1 => (0x00007ffd98751000) libltdl.so.7 => /lib64/libltdl.so.7 (0x00007f930d78b000) libicuuc.so.50 => /lib64/libicuuc.so.50 (0x00007f930d412000) libicudata.so.50 => /lib64/libicudata.so.50 (0x00007f930be3d000) libsasl2.so.3 => /lib64/libsasl2.so.3 (0x00007f930bc20000) libssl.so.10 => /lib64/libssl.so.10 (0x00007f930b9ae000) libcrypto.so.10 => /lib64/libcrypto.so.10 (0x00007f930b54c000) libresolv.so.2 => /lib64/libresolv.so.2 (0x00007f930b332000) libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f930b116000) libc.so.6 => /lib64/libc.so.6 (0x00007f930ad52000) libdl.so.2 => /lib64/libdl.so.2 (0x00007f930ab4e000) libstdc++.so.6 => /lib64/libstdc++.so.6 (0x00007f930a846000) libm.so.6 => /lib64/libm.so.6 (0x00007f930a543000) libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f930a32d000) libcrypt.so.1 => /lib64/libcrypt.so.1 (0x00007f930a0f6000) libgssapi_krb5.so.2 => /lib64/libgssapi_krb5.so.2 (0x00007f9309ea8000) libkrb5.so.3 => /lib64/libkrb5.so.3 (0x00007f9309bc0000) libcom_err.so.2 => /lib64/libcom_err.so.2 (0x00007f93099bc000) libk5crypto.so.3 => /lib64/libk5crypto.so.3 (0x00007f9309788000) libz.so.1 => /lib64/libz.so.1 (0x00007f9309572000) /lib64/ld-linux-x86-64.so.2 (0x000055eefaf24000) libfreebl3.so => /lib64/libfreebl3.so (0x00007f930936e000) libkrb5support.so.0 => /lib64/libkrb5support.so.0 (0x00007f9309160000) libkeyutils.so.1 => /lib64/libkeyutils.so.1 (0x00007f9308f5c000) libselinux.so.1 => /lib64/libselinux.so.1 (0x00007f9308d34000) libpcre.so.1 => /lib64/libpcre.so.1 (0x00007f9308ad2000) Any advice or suggestions would be welcome. Regards, Scott

1 0

Re: Upgrade from 2.4.40 to 2.4.44
by Quanah Gibson-Mount 11 Oct '17

11 Oct '17

--On Friday, September 15, 2017 9:18 AM -0700 Ryan Tandy <ryan(a)nardis.ca> wrote: > IIRC slapcat doesn't work in this case, because it fails to initialize > the ppolicy module. > > The linked CentOS and RHEL bugs recommend downgrading slapd to the > previously working version and using ldapmodify. Yeah, that's ugly :/ Another reason we really need to get slapmodify out, and some way to execute it with an option to not load modules or similar. --Quanah -- Quanah Gibson-Mount Product Architect Symas Corporation Packaged, certified, and supported LDAP solutions powered by OpenLDAP: <http://www.symas.com>

4 3

Replication error
by Ervin Hegedüs 10 Oct '17

10 Oct '17

Hi, I (think I) setting up completly a master-slave replication. The replication user can access from the slave (ldapsearch works). Here is the config, what I added on slave: dn: olcDatabase={1}mdb,cn=config changetype: modify replace: olcSyncRepl olcSyncRepl: rid=001 provider=ldaps://master:636/ bindmethod=simple binddn="uid=repuser,dc=my,dc=domain,dc=hu" credentials=SECRET searchbase="dc=my,dc=domain,dc=hu" scope=sub schemachecking=on type=refreshAndPersist retry="30 5 300 3" interval=00:00:05:00 tls_cacert=/etc/ldap/CAcert.pem tls_cert=/etc/ldap/slave_cert.pem tls_key=/etc/ldap/slave_key.pem tls_reqcert=demand And now I found these lines in syslog: Oct 10 17:36:40 open-ldap2 slapd[4640]: Entry (cn=admin,dc=my,dc=domain,dc=hu): object class 'simpleSecurityObject' requires attribute 'userPassword' Oct 10 17:36:40 open-ldap2 slapd[4640]: null_callback : error code 0x41 Oct 10 17:36:40 open-ldap2 slapd[4640]: syncrepl_entry: rid=001 be_add cn=admin,dc=my,dc=domain,dc=hu failed (65) Oct 10 17:36:41 open-ldap2 slapd[4640]: do_syncrepl: rid=001 rc 65 retrying (4 retries left) I think this occures, because the cn=admin,dc=... user is a simpleSecurityObject, and could't access the userPassword from the ldapsearch - or not :). Anyway - how can I solve this problem? Thanks, a.

1 0

Preauth Issue with SASL-Passthrough and Kerberos Backend
by Ulrich Tehrani 06 Oct '17

06 Oct '17

Hi all, i setup an openldap server which is used as MIT-Kerebros backend. User-Principals have - besides the kerberos attributes - appropriate objectclasses (e.g. simplesecurityObject, organizationalRole) to make also a simple authentication with the attribut userpassword possible. To consolidate both credentials i made a setup of SASL-Pasthrough with backend Kerberos. So i set the value of the userpassword attribut to. {SASL}<user>@<realm> and made the required configurations for the saslauthd. With this configuration all kind of authentications will use the kerberos-password. I made various tests but there seems to be an issue with preauthentication in openldap. I got the follwoing result: =>testsaslauthd is always working if the preauth flag is on or off =>ldapsearch -x is only working with preauth-flag disabled. => It makes no difference if MIT Kerberos use its normal backend Keep in mind: For clear testing condtions saslauthd-caching has to be disabled ! Don't use the -c Option in saslauthd - otherwise it could happen that your ldapsearch -x is working because you had success with a former testsaslauthd-command ! Has someone a similar setup which is working with enabled preauth ? Or does someone know if this is supported or not ? I use LDAP 2.4.44 with cyrus-sasl-2.1.23. Thanks in advance. Kind regards Uli -- =================================== Ulrich Tehrani Am Ulrichshof 19 79189 Bad Krozingen

2 1

Re: Can't get lmdb to build on FreeBSD 12
by Howard Chu 04 Oct '17

04 Oct '17

Russell Haley wrote: > Hi, > > Sorry for the silly question, but I've stumped myself and my C is > weak. Normally lmdb builds fine without issue on most of the platforms > I try (Debian Jessie, FreeBSD 10.3 & 12), but it's stopped compiling > on my TrueOS box, which is FreeBSD 12. I've tried gcc and clang 4.0 > and 3.9. Try using GNU make. It looks like whatever make you're using doesn't have the proper default build rules. > ------------------------ > > russellh@prescott:~/Git/lmdb/libraries/liblmdb% make > cc -pthread -O2 -g -W -Wall -Wno-unused-parameter -Wbad-function-cast > -Wuninitialized -v mdb_stat.c -o mdb_stat > FreeBSD clang version 4.0.0 (tags/RELEASE_400/final 297347) (based on > LLVM 4.0.0) > Target: x86_64-unknown-freebsd12.0 > Thread model: posix > InstalledDir: /usr/bin > "/usr/bin/cc" -cc1 -triple x86_64-unknown-freebsd12.0 -emit-obj > -disable-free -main-file-name mdb_stat.c -mrelocation-model static > -mthread-model posix -mdisable-fp-elim -masm-verbose > -mconstructor-aliases -munwind-tables -target-cpu x86-64 -v > -dwarf-column-info -debug-info-kind=standalone -dwarf-version=2 > -debugger-tuning=gdb -resource-dir /usr/bin/../lib/clang/4.0.0 -O2 -W > -Wall -Wno-unused-parameter -Wbad-function-cast -Wuninitialized > -fdebug-compilation-dir /usr/home/russellh/Git/lmdb/libraries/liblmdb > -ferror-limit 19 -fmessage-length 237 -pthread -fobjc-runtime=gnustep > -fdiagnostics-show-option -vectorize-loops -vectorize-slp -o > /tmp/mdb_stat-852384.o -x c mdb_stat.c > clang -cc1 version 4.0.0 based upon LLVM 4.0.0 default target > x86_64-unknown-freebsd12.0 > #include "..." search starts here: > #include <...> search starts here: > /usr/bin/../lib/clang/4.0.0/include > /usr/include > End of search list. > "/usr/bin/ld" --eh-frame-hdr -dynamic-linker /libexec/ld-elf.so.1 > --hash-style=both --enable-new-dtags -o mdb_stat /usr/lib/crt1.o > /usr/lib/crti.o /usr/lib/crtbegin.o -L/usr/lib /tmp/mdb_stat-852384.o > -lgcc --as-needed -lgcc_s --no-as-needed -lpthread -lc -lgcc > --as-needed -lgcc_s --no-as-needed /usr/lib/crtend.o /usr/lib/crtn.o > /tmp/mdb_stat-852384.o: In function `main': > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:104: > undefined reference to `mdb_env_create' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:106: > undefined reference to `mdb_strerror' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:111: > undefined reference to `mdb_env_set_maxdbs' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:114: > undefined reference to `mdb_env_open' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:116: > undefined reference to `mdb_strerror' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:121: > undefined reference to `mdb_env_stat' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:122: > undefined reference to `mdb_env_info' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:136: > undefined reference to `mdb_reader_list' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:139: > undefined reference to `mdb_reader_check' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:141: > undefined reference to `mdb_reader_list' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:147: > undefined reference to `mdb_txn_begin' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:149: > undefined reference to `mdb_strerror' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:257: > undefined reference to `mdb_env_close' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:160: > undefined reference to `mdb_cursor_open' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:162: > undefined reference to `mdb_strerror' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:165: > undefined reference to `mdb_stat' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:171: > undefined reference to `mdb_cursor_get' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:171: > undefined reference to `mdb_cursor_get' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:199: > undefined reference to `mdb_cursor_close' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:203: > undefined reference to `mdb_dbi_open' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:205: > undefined reference to `mdb_strerror' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:209: > undefined reference to `mdb_stat' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:(.text+0x5d6): > undefined reference to `mdb_strerror' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:255: > undefined reference to `mdb_txn_abort' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:221: > undefined reference to `mdb_cursor_open' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:223: > undefined reference to `mdb_strerror' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:226: > undefined reference to `mdb_cursor_get' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:234: > undefined reference to `mdb_dbi_open' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:239: > undefined reference to `mdb_stat' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:245: > undefined reference to `mdb_dbi_close' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:226: > undefined reference to `mdb_cursor_get' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:247: > undefined reference to `mdb_cursor_close' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:253: > undefined reference to `mdb_dbi_close' > /usr/home/russellh/Git/lmdb/libraries/liblmdb/mdb_stat.c:241: > undefined reference to `mdb_strerror' > cc: error: linker command failed with exit code 1 (use -v to see > invocation) > *** Error code 1 > > Stop. > make: stopped in /usr/home/russellh/Git/lmdb/libraries/liblmdb > > ---------------------------------- > The full output is here with uname, git status and compiler info: > http://termbin.com/io9d > > Any help would be grand. > > Thanks, > Russ > > -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

3 2

Re: lmdb as a Lua database
by Russell Haley 04 Oct '17

04 Oct '17

Brilliant. May I ping you for more information? I'm tightly coupled to my current toolchain but will use this for reference. Russ On Wed, Oct 4, 2017 at 9:21 AM, Леонид Юрьев <leo(a)yuriev.ru> wrote: > take a look at https://github.com/PositiveTechnologies/libfpta > > regads. > > > 2017-10-04 9:20 GMT+03:00 Russell Haley <russ.haley(a)gmail.com>: >> On Tue, Oct 3, 2017 at 11:19 PM, Russell Haley <russ.haley(a)gmail.com> wrote: >>> Hi, >>> >>> So I'm using LMDB to create a simple 'database' system for lua that >>> will allow me to directly serialize and store lua table data. So far I >>> have used the wonderful lmdb wrapper lightningmdb and created a simple >>> API to hide some of the mechanisms. >>> >>> Honestly, your API is so well thought out, I'm doing a disservice by >>> removing a lot of the powerful features. However, I press ahead and >>> beg forgiveness. :) >>> >>> I have two questions: >>> >>> 1) There is a simplified vs cursor API. I am starting to suspect that >>> the simplified functions are just basic wrappers around the cursor >>> API? I ask because if this is the case, I could just remove my support >>> for the simplified API and only have to support one set of functions. >>> >>> 2) The key lookup is blazing fast (so awesome, this thing is going to >>> smoke!). I wanted to create a way of indexing the data elements within >>> the database by allowing users to create lua functions that parse the >>> lua table values. >>> --a basic table that would be serialized into a lmdb database called t1 >>> t1 = { >>> 1={col_a="test",col_b="item1"}, >>> 2={col_a="not-a-test",col_b="item2"}, >>> 3={col_a="test",col_b="item3"} >>> } >>> >>> --The function that would be called per-row to create an index. >>> local function(k,v) >>> >>> return v.col_a, k >>> end >>> >>> The idea would be to use the dupsort feature to create an 'index' >>> database with data structured as: >>> >>> not-a-test, 2 >>> test, 1 >>> test, 3 >> >> Sorry, hit send in gmail by accident. Anyway, the idea is to use the >> dupsort feature to create indexes. Is this feasible? Any input or >> direction would be fantastic. >> >> Thanks! >> Russ >>

1 0

Re: lmdb as a Lua database
by Howard Chu 04 Oct '17

04 Oct '17

Russell Haley wrote: > Hi, > > So I'm using LMDB to create a simple 'database' system for lua that > will allow me to directly serialize and store lua table data. So far I > have used the wonderful lmdb wrapper lightningmdb and created a simple > API to hide some of the mechanisms. > > Honestly, your API is so well thought out, I'm doing a disservice by > removing a lot of the powerful features. However, I press ahead and > beg forgiveness. :) > > I have two questions: > > 1) There is a simplified vs cursor API. I am starting to suspect that > the simplified functions are just basic wrappers around the cursor > API? I ask because if this is the case, I could just remove my support > for the simplified API and only have to support one set of functions. Yes, everything uses cursors internally. > 2) The key lookup is blazing fast (so awesome, this thing is going to > smoke!). I wanted to create a way of indexing the data elements within > the database by allowing users to create lua functions that parse the > lua table values. > --a basic table that would be serialized into a lmdb database called t1 > t1 = { > 1={col_a="test",col_b="item1"}, > 2={col_a="not-a-test",col_b="item2"}, > 3={col_a="test",col_b="item3"} > } > > --The function that would be called per-row to create an index. > local function(k,v) > > return v.col_a, k > end > > The idea would be to use the dupsort feature to create an 'index' > database with data structured as: > > not-a-test, 2 > test, 1 > test, 3 Sure, that's generally what dupsort is for. -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

Re: Initialize ldap with mdb
by Howard Chu 04 Oct '17

04 Oct '17

rammohan ganapavarapu wrote: > Howard, > > If it doesnt swap, what will it do in case of db size is larger than actual > RAM? also how does it manage the physical RAM? meaning what % system RAM does > it use? It doesn't manage physical RAM at all, the OS does. There is no fixed % of system RAM that it will use, the OS will let it use whatever is available. > is it configurable ? No. > will it case any OOM issues? No. > i am trying to > migrate from hdb to mdb in prod so trying to understand mdb. How does it work > in case of reads and writes? > > Thanks, > Ram > > On Tue, Oct 3, 2017 at 4:15 PM, Howard Chu <hyc(a)symas.com > <mailto:hyc@symas.com>> wrote: > > Quanah Gibson-Mount wrote: > > --On Tuesday, October 03, 2017 11:47 AM -0700 rammohan ganapavarapu > <rammohanganap(a)gmail.com <mailto:rammohanganap@gmail.com>> wrote: > > > Ok, but are there any casses where DB size can grow bigger then > RAM and > what will happen in that case? > > > Same thing that happens in any such case.. it'll start swapping. > > > LMDB never uses swap. > > -- > -- Howard Chu > CTO, Symas Corp. http://www.symas.com > Director, Highland Sun http://highlandsun.com/hyc/ > Chief Architect, OpenLDAP http://www.openldap.org/project/ > <http://www.openldap.org/project/> > > -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

openldap-technical October 2017